P-value correction: Difference between revisions
→{{header|Kotlin}}: Added second version to deal with licensing issues of first version. |
→{{header|C}}: Added second version to avoid licensing issues of first version. |
||
Line 26:
=={{header|C}}==
===Version 1===
{{works with|C99}}
{{trans|R}}
Line 677 ⟶ 678:
type 5 = 'hommel' has cumulative error of 4.35302e-07</pre>
===Version 2===
{{works with|C89}}
{{trans|Kotlin}}
To avoid licensing issues, this version is a translation of the Kotlin entry (Version 2) which is itself a partial translation of the Perl 6 entry. If using gcc, you need to link to the math library (-lm).
<lang C>#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#define SIZE 50
#define each_i(start, end) for (i = start; i < end; ++i)
typedef enum { UP, DOWN } direction;
typedef struct { int index; double value; } iv1;
typedef struct { int index; int value; } iv2;
/* test also for 'Unknown' correction type */
const char *types[8] = {
"Benjamini-Hochberg", "Benjamini-Yekutieli", "Bonferroni", "Hochberg",
"Holm", "Hommel", "Šidák", "Unknown"
};
int compare_iv1(const void *a, const void *b) {
double aa = ((iv1 *)a) -> value;
double bb = ((iv1 *)b) -> value;
if (aa > bb) return 1;
if (aa < bb) return -1;
return 0;
}
int compare_iv1_desc(const void *a, const void *b) {
return -compare_iv1(a, b);
}
int compare_iv2(const void *a, const void *b) {
return ((iv2 *)a) -> value - ((iv2 *)b) -> value;
}
void ratchet(double *pa, direction dir) {
int i;
double m = pa[0];
if (dir == UP) {
each_i(1, SIZE) {
if (pa[i] > m) pa[i] = m;
m = pa[i];
}
}
else {
each_i(1, SIZE) {
if (pa[i] < m) pa[i] = m;
m = pa[i];
}
}
each_i(0, SIZE) if (pa[i] > 1.0) pa[i] = 1.0;
}
void schwartzian(const double *p, double *pa, direction dir) {
int i;
int order[SIZE];
int order2[SIZE];
iv1 iv1s[SIZE];
iv2 iv2s[SIZE];
double pa2[SIZE];
each_i(0, SIZE) { iv1s[i].index = i; iv1s[i].value = p[i]; }
if (dir == UP)
qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1_desc);
else
qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1);
each_i(0, SIZE) order[i] = iv1s[i].index;
each_i(0, SIZE) pa[i] *= p[order[i]];
ratchet(pa, dir);
each_i(0, SIZE) { iv2s[i].index = i; iv2s[i].value = order[i]; }
qsort(iv2s, SIZE, sizeof(iv2s[0]), compare_iv2);
each_i(0, SIZE) order2[i] = iv2s[i].index;
each_i(0, SIZE) pa2[i] = pa[order2[i]];
each_i(0, SIZE) pa[i] = pa2[i];
}
void adjust(const double *p, double *pa, const char *type) {
int i;
if (!strcmp(type, "Benjamini-Hochberg")) {
each_i(0, SIZE) pa[i] = (double)SIZE / (SIZE - i);
schwartzian(p, pa, UP);
}
else if (!strcmp(type, "Benjamini-Yekutieli")) {
double q = 0.0;
each_i(1, SIZE + 1) q += 1.0 / i;
each_i(0, SIZE) pa[i] = q * SIZE / (SIZE - i);
schwartzian(p, pa, UP);
}
else if (!strcmp(type, "Bonferroni")) {
each_i(0, SIZE) pa[i] = (p[i] * SIZE > 1.0) ? 1.0 : p[i] * SIZE;
}
else if (!strcmp(type, "Hochberg")) {
each_i(0, SIZE) pa[i] = i + 1.0;
schwartzian(p, pa, UP);
}
else if (!strcmp(type, "Holm")) {
each_i(0, SIZE) pa[i] = SIZE - i;
schwartzian(p, pa, DOWN);
}
else if (!strcmp(type, "Hommel")) {
int i, j;
int order[SIZE];
int order2[SIZE];
iv1 iv1s[SIZE];
iv2 iv2s[SIZE];
double s[SIZE];
double q[SIZE];
double pa2[SIZE];
int indices[SIZE];
each_i(0, SIZE) { iv1s[i].index = i; iv1s[i].value = p[i]; }
qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1);
each_i(0, SIZE) order[i] = iv1s[i].index;
each_i(0, SIZE) s[i] = p[order[i]];
double min = s[0] * SIZE;
each_i(1, SIZE) {
double temp = s[i] / (i + 1.0);
if (temp < min) min = temp;
}
each_i(0, SIZE) q[i] = min;
each_i(0, SIZE) pa2[i] = min;
for (j = SIZE - 1; j >= 2; --j) {
each_i(0, SIZE) indices[i] = i;
int upper_start = SIZE - j + 1; /* upper indices start index */
int upper_size = j - 1; /* size of upper indices */
int lower_size = SIZE - upper_size; /* size of lower indices */
double qmin = j * s[indices[upper_start]] / 2.0;
each_i(1, upper_size) {
double temp = s[indices[upper_start + i]] * j / (2.0 + i);
if (temp < qmin) qmin = temp;
}
each_i(0, lower_size) {
double temp = s[indices[i]] * j;
q[indices[i]] = (temp < qmin) ? temp : qmin;
}
each_i(0, upper_size) q[indices[upper_start + i]] = q[SIZE - j];
each_i(0, SIZE) if (pa2[i] < q[i]) pa2[i] = q[i];
}
each_i(0, SIZE) { iv2s[i].index = i; iv2s[i].value = order[i]; }
qsort(iv2s, SIZE, sizeof(iv2s[0]), compare_iv2);
each_i(0, SIZE) order2[i] = iv2s[i].index;
each_i(0, SIZE) pa[i] = pa2[order2[i]];
}
else if (!strcmp(type, "Šidák")) {
each_i(0, SIZE) pa[i] = 1.0 - pow(1.0 - p[i], SIZE);
}
else {
printf("\nSorry, do not know how to do '%s' correction.\n", type);
printf("Perhaps you want one of these?:\n");
each_i(0, 7) printf(" %s\n", types[i]);
exit(1);
}
}
void adjusted(const double *p, const char *type) {
int i;
double pa[SIZE] = { 0.0 };
if (check(p)) {
adjust(p, pa, type);
printf("\n%s", type);
each_i(0, SIZE) {
if (!(i % 5)) printf("\n[%2d] ", i);
printf("%1.10f ", pa[i]);
}
printf("\n");
}
else {
printf("p-values must be in range 0.0 to 1.0\n");
exit(1);
}
}
int check(const double* p) {
int i;
each_i(0, SIZE) {
if (p[i] < 0.0 || p[i] > 1.0) return 0;
}
return 1;
}
int main() {
int i;
double p_values[SIZE] = {
4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01,
8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01,
4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01,
8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02,
3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01,
1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02,
4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04,
3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04,
1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04,
2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03
};
each_i(0, 8) adjusted(p_values, types[i]);
return 0;
}</lang>
{{output}}
<pre>
Same as Kotlin (Version 2) output.
</pre>
=={{header|D}}==
|
Revision as of 17:30, 20 March 2018
Given a list of p-values, adjust the p-values for multiple comparisons. This is done in order to control the false positive, or Type 1 error rate. This is also known as the "false discovery rate" (FDR). After adjustment, the p-values will be higher but still inside [0,1]. The adjusted p-values are sometimes called "q-values".
- Task
Given one list of p-values, return the p-values correcting for multiple comparisons
p = {4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03}
There are several methods to do this, see:
- Yoav Benjamini, Yosef Hochberg "Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing", Journal of the Royal Statistical Society. Series B, Vol. 57, No. 1 (1995), pp. 289-300, JSTOR:2346101
- Yoav Benjamini, Daniel Yekutieli, "The control of the false discovery rate in multiple testing under dependency", Ann. Statist., Vol. 29, No. 4 (2001), pp. 1165-1188, DOI:10.1214/aos/1013699998 JSTOR:2674075
- Sture Holm, "A Simple Sequentially Rejective Multiple Test Procedure", Scandinavian Journal of Statistics, Vol. 6, No. 2 (1979), pp. 65-70, JSTOR:4615733
- Yosef Hochberg, "A sharper Bonferroni procedure for multiple tests of significance", Biometrika, Vol. 75, No. 4 (1988), pp 800–802, DOI:10.1093/biomet/75.4.800 JSTOR:2336325
- Gerhard Hommel, "A stagewise rejective multiple test procedure based on a modified Bonferroni test", Biometrika, Vol. 75, No. 2 (1988), pp 383–386, DOI:10.1093/biomet/75.2.383 JSTOR:2336190
Each method has its own advantages and disadvantages.
C
Version 1
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses.
This work is a translation of the R source code. In order to confirm that the new function is working correctly, each value is compared to R's output and a cumulative absolute error is returned.
The C function p_adjust
is designed to work as similarly to the R function p.adjust
as possible, and is able to do any one of the methods.
This program, for example, fdr.c, can be compiled by
gcc -o fdr fdr.c -Wall -pedantic -std=c11 -lm -O4
or
clang -o fdr fdr.c -Wall -pedantic -std=c11 -lm -O4
.
Link with -lm
<lang C>#include <stdio.h>//printf
- include <stdlib.h>//qsort
- include <math.h>//fabs
- include <stdbool.h>//bool data type
- include <strings.h>//strcasecmp
unsigned int *restrict seq_len(const size_t START, const size_t END) { //named after R function of same name, but simpler function size_t start = START; size_t end = END; if (START == END) { unsigned int *restrict sequence = malloc( (end+1) * sizeof(unsigned int)); if (sequence == NULL) { printf("malloc failed at %s line %u\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (size_t i = 0; i < end; i++) { sequence[i] = i+1; } return sequence; } if (START > END) { end = START; start = END; } const size_t LENGTH = end - start ; unsigned int *restrict sequence = malloc( (1+LENGTH) * sizeof(unsigned int)); if (sequence == NULL) { printf("malloc failed at %s line %u\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } if (START < END) { for (size_t index = 0; index <= LENGTH; index++) { sequence[index] = start + index; } } else { for (size_t index = 0; index <= LENGTH; index++) { sequence[index] = end - index; } } return sequence; }
//modified from https://phoxis.org/2012/07/12/get-sorted-index-orderting-of-an-array/
double *restrict base_arr = NULL;
static int compar_increase (const void *restrict a, const void *restrict b) { int aa = *((int *restrict ) a), bb = *((int *restrict) b); if (base_arr[aa] < base_arr[bb]) { return 1; } else if (base_arr[aa] == base_arr[bb]) { return 0; } else { return -1; } }
static int compar_decrease (const void *restrict a, const void *restrict b) { int aa = *((int *restrict ) a), bb = *((int *restrict) b); if (base_arr[aa] < base_arr[bb]) { return -1; } else if (base_arr[aa] == base_arr[bb]) { return 0; } else { return 1; } }
unsigned int *restrict order (const double *restrict ARRAY, const unsigned int SIZE, const bool DECREASING) { //this has the same name as the same R function unsigned int *restrict idx = malloc(SIZE * sizeof(unsigned int)); if (idx == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } base_arr = malloc(sizeof(double) * SIZE); if (base_arr == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (unsigned int i = 0; i < SIZE; i++) { base_arr[i] = ARRAY[i]; idx[i] = i; } if (DECREASING == false) { qsort(idx, SIZE, sizeof(unsigned int), compar_decrease); } else if (DECREASING == true) { qsort(idx, SIZE, sizeof(unsigned int), compar_increase); } free(base_arr); base_arr = NULL; return idx; }
double *restrict cummin(const double *restrict ARRAY, const unsigned int NO_OF_ARRAY_ELEMENTS) { //this takes the same name of the R function which it copies //this requires a free() afterward where it is used if (NO_OF_ARRAY_ELEMENTS < 1) { puts("cummin function requires at least one element.\n"); printf("Failed at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } double *restrict output = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (output == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } double cumulative_min = ARRAY[0]; for (unsigned int i = 0; i < NO_OF_ARRAY_ELEMENTS; i++) { if (ARRAY[i] < cumulative_min) { cumulative_min = ARRAY[i]; } output[i] = cumulative_min; } return output; }
double *restrict cummax(const double *restrict ARRAY, const unsigned int NO_OF_ARRAY_ELEMENTS) { //this takes the same name of the R function which it copies //this requires a free() afterward where it is used if (NO_OF_ARRAY_ELEMENTS < 1) { puts("function requires at least one element.\n"); printf("Failed at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } double *restrict output = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (output == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } double cumulative_max = ARRAY[0]; for (size_t i = 0; i < NO_OF_ARRAY_ELEMENTS; i++) { if (ARRAY[i] > cumulative_max) { cumulative_max = ARRAY[i]; } output[i] = cumulative_max; } return output; }
double *restrict pminx(const double *restrict ARRAY, const size_t NO_OF_ARRAY_ELEMENTS, const double X) { //named after the R function pmin if (NO_OF_ARRAY_ELEMENTS < 1) { puts("pmin requires at least one element.\n"); printf("Failed at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } double *restrict pmin_array = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (pmin_array == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (unsigned int index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { if (ARRAY[index] < X) { pmin_array[index] = ARRAY[index]; } else { pmin_array[index] = X; } } return pmin_array; }
void double_say (const double *restrict ARRAY, const size_t NO_OF_ARRAY_ELEMENTS) { printf("[1] %e", ARRAY[0]); for (size_t i = 1; i < NO_OF_ARRAY_ELEMENTS; i++) { printf(" %.10f", ARRAY[i]); if (((i+1) % 5) == 0) { printf("\n[%zu]", i+1); } } puts("\n"); }
/*void uint_say (const unsigned int *restrict ARRAY, const size_t NO_OF_ARRAY_ELEMENTS) { //for debugging printf("%u", ARRAY[0]); for (size_t i = 1; i < NO_OF_ARRAY_ELEMENTS; i++) { printf(",%u", ARRAY[i]); } puts("\n"); }*/
double *restrict uint2double (const unsigned int *restrict ARRAY, const size_t NO_OF_ARRAY_ELEMENTS) { double *restrict doubleArray = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (doubleArray == NULL) { printf("Failure to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { doubleArray[index] = (double)ARRAY[index]; } return doubleArray; }
double min2 (const double N1, const double N2) { if (N1 < N2) { return N1; } else { return N2; } }
double *restrict p_adjust (const double *restrict PVALUES, const size_t NO_OF_ARRAY_ELEMENTS, const char *restrict STRING) { //this function is a translation of R's p.adjust "BH" method // i is always i[index] = NO_OF_ARRAY_ELEMENTS - index - 1 if (NO_OF_ARRAY_ELEMENTS < 1) { puts("p_adjust requires at least one element.\n"); printf("Failed at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } short int TYPE = -1; if (strcasecmp(STRING, "BH") == 0) { TYPE = 0; } else if (strcasecmp(STRING, "fdr") == 0) { TYPE = 0; } else if (strcasecmp(STRING, "by") == 0) { TYPE = 1; } else if (strcasecmp(STRING, "Bonferroni") == 0) { TYPE = 2; } else if (strcasecmp(STRING, "hochberg") == 0) { TYPE = 3; } else if (strcasecmp(STRING, "holm") == 0) { TYPE = 4; } else if (strcasecmp(STRING, "hommel") == 0) { TYPE = 5; } else { printf("%s doesn't match any accepted FDR methods.\n", STRING); printf("Failed at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- if (TYPE == 2) {//Bonferroni method double *restrict bonferroni = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (bonferroni == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { const double BONFERRONI = PVALUES[index] * NO_OF_ARRAY_ELEMENTS; if (BONFERRONI >= 1.0) { bonferroni[index] = 1.0; } else if ((0.0 <= BONFERRONI) && (BONFERRONI < 1.0)) { bonferroni[index] = BONFERRONI; } else { printf("%g is outside of the interval I planned.\n", BONFERRONI); printf("Failure at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } } return bonferroni; //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- } else if (TYPE == 4) {//Holm method /*these values are computed separately from BH, BY, and Hochberg because they are computed differently*/ unsigned int *restrict o = order(PVALUES, NO_OF_ARRAY_ELEMENTS, false); //sorted in reverse of methods 0-3 double *restrict o2double = uint2double(o, NO_OF_ARRAY_ELEMENTS); double *restrict cummax_input = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { cummax_input[index] = (NO_OF_ARRAY_ELEMENTS - index ) * PVALUES[o[index]]; // printf("cummax_input[%zu] = %e\n", index, cummax_input[index]); } free(o); o = NULL; unsigned int *restrict ro = order(o2double, NO_OF_ARRAY_ELEMENTS, false); free(o2double); o2double = NULL;
double *restrict cummax_output = cummax(cummax_input, NO_OF_ARRAY_ELEMENTS); free(cummax_input); cummax_input = NULL;
double *restrict pmin = pminx(cummax_output, NO_OF_ARRAY_ELEMENTS, 1); free(cummax_output); cummax_output = NULL; double *restrict qvalues = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { qvalues[index] = pmin[ro[index]]; } free(pmin); pmin = NULL; free(ro); ro = NULL; return qvalues; //--------------------------------------------------------------------------- //--------------------------------------------------------------------------- } else if (TYPE == 5) {//Hommel method //i <- seq_len(n) //o <- order(p) unsigned int *restrict o = order(PVALUES, NO_OF_ARRAY_ELEMENTS, false);//false is R's default //p <- p[o] double *restrict p = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (p == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { p[index] = PVALUES[o[index]]; } //ro <- order(o) double *restrict o2double = uint2double(o, NO_OF_ARRAY_ELEMENTS); free(o); o = NULL; unsigned int *restrict ro = order(o2double, NO_OF_ARRAY_ELEMENTS, false); free(o2double); o2double = NULL; // puts("ro"); //q <- pa <- rep.int(min(n * p/i), n) double *restrict q = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (q == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } double *restrict pa = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (pa == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } double min = (double)NO_OF_ARRAY_ELEMENTS * p[0]; for (size_t index = 1; index < NO_OF_ARRAY_ELEMENTS; index++) { const double TEMP = (double)NO_OF_ARRAY_ELEMENTS * p[index] / (1+index); if (TEMP < min) { min = TEMP; } } for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { pa[index] = min; q[index] = min; } // puts("q & pa"); // double_say(q, NO_OF_ARRAY_ELEMENTS); /*for (j in (n - 1):2) {
ij <- seq_len(n - j + 1) i2 <- (n - j + 2):n q1 <- min(j * p[i2]/(2:j)) q[ij] <- pmin(j * p[ij], q1) q[i2] <- q[n - j + 1] pa <- pmax(pa, q) }
- /
for (size_t j = (NO_OF_ARRAY_ELEMENTS-1); j >= 2; j--) { // printf("j = %zu\n", j); unsigned int *restrict ij = seq_len(1,NO_OF_ARRAY_ELEMENTS - j + 1); for (size_t i = 0; i < NO_OF_ARRAY_ELEMENTS - j + 1; i++) { ij[i]--;//R's indices are 1-based, C's are 0-based } const size_t I2_LENGTH = j - 1; unsigned int *restrict i2 = malloc(I2_LENGTH * sizeof(unsigned int)); for (size_t i = 0; i < I2_LENGTH; i++) { i2[i] = NO_OF_ARRAY_ELEMENTS-j+2+i-1; //R's indices are 1-based, C's are 0-based, I added the -1 }
double q1 = j * p[i2[0]] / 2.0; for (size_t i = 1; i < I2_LENGTH; i++) {//loop through 2:j const double TEMP_Q1 = (double)j * p[i2[i]] / (2 + i); if (TEMP_Q1 < q1) { q1 = TEMP_Q1; } }
for (size_t i = 0; i < (NO_OF_ARRAY_ELEMENTS - j + 1); i++) {//q[ij] <- pmin(j * p[ij], q1) q[ij[i]] = min2( j*p[ij[i]], q1); } free(ij); ij = NULL;
for (size_t i = 0; i < I2_LENGTH; i++) {//q[i2] <- q[n - j + 1] q[i2[i]] = q[NO_OF_ARRAY_ELEMENTS - j];//subtract 1 because of starting index difference } free(i2); i2 = NULL;
for (size_t i = 0; i < NO_OF_ARRAY_ELEMENTS; i++) {//pa <- pmax(pa, q) if (pa[i] < q[i]) { pa[i] = q[i]; } } // printf("j = %zu, pa = \n", j); // double_say(pa, N); }//end j loop free(p); p = NULL; for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { q[index] = pa[ro[index]];//Hommel q-values } //now free memory free(ro); ro = NULL; free(pa); pa = NULL; return q; } //The methods are similarly computed and thus can be combined for clarity unsigned int *restrict o = order(PVALUES, NO_OF_ARRAY_ELEMENTS, true); if (o == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } double *restrict o_double = uint2double(o, NO_OF_ARRAY_ELEMENTS); for (unsigned int index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { if ((PVALUES[index] < 0) || (PVALUES[index] > 1)) { printf("array[%u] = %lf, which is outside the interval [0,1]\n", index, PVALUES[index]); printf("died at %s line %u\n", __FILE__, __LINE__); exit(EXIT_FAILURE); } }
unsigned int *restrict ro = order(o_double, NO_OF_ARRAY_ELEMENTS, false); if (ro == NULL) { printf("failed to malloc at %s line %u.\n", __FILE__, __LINE__); perror(""); exit(EXIT_FAILURE); } free(o_double); o_double = NULL; double *restrict cummin_input = malloc(sizeof(double) * NO_OF_ARRAY_ELEMENTS); if (TYPE == 0) {//BH method for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { const double NI = (double)NO_OF_ARRAY_ELEMENTS / (NO_OF_ARRAY_ELEMENTS - index);// n/i simplified cummin_input[index] = NI * PVALUES[o[index]];//PVALUES[o[index]] is p[o] } } else if (TYPE == 1) {//BY method double q = 1.0; for (size_t index = 2; index < (1+NO_OF_ARRAY_ELEMENTS); index++) { q += (double) 1.0/index; } for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { const double NI = (double)NO_OF_ARRAY_ELEMENTS / (NO_OF_ARRAY_ELEMENTS - index);// n/i simplified cummin_input[index] = q * NI * PVALUES[o[index]];//PVALUES[o[index]] is p[o] } } else if (TYPE == 3) {//Hochberg method for (size_t index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { // pmin(1, cummin((n - i + 1L) * p[o]))[ro] cummin_input[index] = (index + 1) * PVALUES[o[index]]; } } free(o); o = NULL; double *restrict cummin_array = NULL; cummin_array = cummin(cummin_input, NO_OF_ARRAY_ELEMENTS); free(cummin_input); cummin_input = NULL;//I don't need this anymore double *restrict pmin = pminx(cummin_array, NO_OF_ARRAY_ELEMENTS, 1); free(cummin_array); cummin_array = NULL; double *restrict q_array = malloc(NO_OF_ARRAY_ELEMENTS*sizeof(double)); for (unsigned int index = 0; index < NO_OF_ARRAY_ELEMENTS; index++) { q_array[index] = pmin[ro[index]]; }
free(ro); ro = NULL; free(pmin); pmin = NULL; return q_array; }
int main(void) { const double PVALUES[] = {4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03};//just the pvalues const double CORRECT_ANSWERS[6][50] = {//each first index is type {6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02},//Benjamini-Hochberg {1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02},//Benjamini & Yekutieli {1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01},//Bonferroni {9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01},//Hochberg {1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01},//Holm { 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01}//Hommel }; //the following loop checks each type with R's answers const char *restrict TYPES[] = {"bh", "by", "bonferroni", "hochberg", "holm", "hommel"}; for (unsigned short int type = 0; type <= 5; type++) { double *restrict q = p_adjust(PVALUES, sizeof(PVALUES) / sizeof(*PVALUES), TYPES[type]); double error = fabs(q[0] - CORRECT_ANSWERS[type][0]); // printf("%e - %e = %g\n", q[0], CORRECT_ANSWERS[type][0], error); // puts("p q"); // printf("%g\t%g\n", pvalues[0], q[0]); for (unsigned int i = 1; i < sizeof(PVALUES) / sizeof(*PVALUES); i++) { const double this_error = fabs(q[i] - CORRECT_ANSWERS[type][i]); // printf("%e - %e = %g\n", q[i], CORRECT_ANSWERS[type][i], error); error += this_error; } double_say(q, sizeof(PVALUES) / sizeof(*PVALUES)); free(q); q = NULL; printf("\ntype %u = '%s' has cumulative error of %g\n", type, TYPES[type], error); }
return 0; } </lang>
- Output:
[1] 6.126681e-01 0.8521710465 0.1987205200 0.1891595417 0.3217789286 [5] 0.9301450000 0.4870370000 0.9301450000 0.6049730556 0.6826752564 [10] 0.6482628947 0.7253722500 0.5280972727 0.8769925556 0.4705703448 [15] 0.9241867391 0.6049730556 0.7856107317 0.4887525806 0.1136717045 [20] 0.4991890625 0.8769925556 0.9991834000 0.3217789286 0.9301450000 [25] 0.2304957692 0.5832475000 0.0389954722 0.8521710465 0.1476842609 [30] 0.0168363750 0.0025629017 0.0351608437 0.0625018947 0.0036365888 [35] 0.0025629017 0.0294688286 0.0061660636 0.0389954722 0.0026889914 [40] 0.0004502862 0.0000125223 0.0788155476 0.0314261300 0.0048465270 [45] 0.0025629017 0.0048465270 0.0011017083 0.0725203250 0.0220595769 [50] type 0 = 'bh' has cumulative error of 8.03053e-07 [1] 1.000000e+00 1.0000000000 0.8940844244 0.8510676197 1.0000000000 [5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 0.5114323399 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.1754486368 1.0000000000 0.6644618149 [30] 0.0757503083 0.0115310209 0.1581958559 0.2812088585 0.0163617595 [35] 0.0115310209 0.1325863108 0.0277423864 0.1754486368 0.0120983246 [40] 0.0020259303 0.0000563403 0.3546073326 0.1413926119 0.0218055202 [45] 0.0115310209 0.0218055202 0.0049568120 0.3262838334 0.0992505663 [50] type 1 = 'by' has cumulative error of 3.64072e-07 [1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.7019185000 1.0000000000 1.0000000000 [30] 0.2020365000 0.0151667450 0.5625735000 1.0000000000 0.0290927100 [35] 0.0153774100 0.4125636000 0.0678267000 0.6803480000 0.0188229400 [40] 0.0009005725 0.0000125223 1.0000000000 0.4713919500 0.0439557650 [45] 0.0108891550 0.0484652700 0.0033051250 1.0000000000 0.2867745000 [50] type 2 = 'bonferroni' has cumulative error of 6.5e-08 [1] 9.991834e-01 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4632662100 0.9991834000 0.9991834000 [30] 0.1575884700 0.0138396690 0.3938014500 0.7600230400 0.0250197306 [35] 0.0138396690 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 3 = 'hochberg' has cumulative error of 2.7375e-07 [1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.4632662100 1.0000000000 1.0000000000 [30] 0.1575884700 0.0139534054 0.3938014500 0.7600230400 0.0250197306 [35] 0.0139534054 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 4 = 'holm' has cumulative error of 2.8095e-07 [1] 9.991834e-01 0.9991834000 0.9991834000 0.9987623800 0.9991834000 [5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9595180000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4351894700 0.9991834000 0.9766522500 [30] 0.1414255500 0.0130434007 0.3530936533 0.6887708800 0.0238560222 [35] 0.0132245726 0.2722919760 0.0542613600 0.4218157600 0.0158112696 [40] 0.0008825610 0.0000125223 0.8743649143 0.3016908480 0.0351646120 [45] 0.0095824564 0.0387722160 0.0031729200 0.8122276400 0.1950066600 [50] type 5 = 'hommel' has cumulative error of 4.35302e-07
Version 2
To avoid licensing issues, this version is a translation of the Kotlin entry (Version 2) which is itself a partial translation of the Perl 6 entry. If using gcc, you need to link to the math library (-lm). <lang C>#include <stdio.h>
- include <stdlib.h>
- include <math.h>
- define SIZE 50
- define each_i(start, end) for (i = start; i < end; ++i)
typedef enum { UP, DOWN } direction;
typedef struct { int index; double value; } iv1;
typedef struct { int index; int value; } iv2;
/* test also for 'Unknown' correction type */ const char *types[8] = {
"Benjamini-Hochberg", "Benjamini-Yekutieli", "Bonferroni", "Hochberg", "Holm", "Hommel", "Šidák", "Unknown"
};
int compare_iv1(const void *a, const void *b) {
double aa = ((iv1 *)a) -> value; double bb = ((iv1 *)b) -> value; if (aa > bb) return 1; if (aa < bb) return -1; return 0;
}
int compare_iv1_desc(const void *a, const void *b) {
return -compare_iv1(a, b);
}
int compare_iv2(const void *a, const void *b) {
return ((iv2 *)a) -> value - ((iv2 *)b) -> value;
}
void ratchet(double *pa, direction dir) {
int i; double m = pa[0]; if (dir == UP) { each_i(1, SIZE) { if (pa[i] > m) pa[i] = m; m = pa[i]; } } else { each_i(1, SIZE) { if (pa[i] < m) pa[i] = m; m = pa[i]; } } each_i(0, SIZE) if (pa[i] > 1.0) pa[i] = 1.0;
}
void schwartzian(const double *p, double *pa, direction dir) {
int i; int order[SIZE]; int order2[SIZE]; iv1 iv1s[SIZE]; iv2 iv2s[SIZE]; double pa2[SIZE]; each_i(0, SIZE) { iv1s[i].index = i; iv1s[i].value = p[i]; } if (dir == UP) qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1_desc); else qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1); each_i(0, SIZE) order[i] = iv1s[i].index; each_i(0, SIZE) pa[i] *= p[order[i]]; ratchet(pa, dir); each_i(0, SIZE) { iv2s[i].index = i; iv2s[i].value = order[i]; } qsort(iv2s, SIZE, sizeof(iv2s[0]), compare_iv2); each_i(0, SIZE) order2[i] = iv2s[i].index; each_i(0, SIZE) pa2[i] = pa[order2[i]]; each_i(0, SIZE) pa[i] = pa2[i];
}
void adjust(const double *p, double *pa, const char *type) {
int i; if (!strcmp(type, "Benjamini-Hochberg")) { each_i(0, SIZE) pa[i] = (double)SIZE / (SIZE - i); schwartzian(p, pa, UP); } else if (!strcmp(type, "Benjamini-Yekutieli")) { double q = 0.0; each_i(1, SIZE + 1) q += 1.0 / i; each_i(0, SIZE) pa[i] = q * SIZE / (SIZE - i); schwartzian(p, pa, UP); } else if (!strcmp(type, "Bonferroni")) { each_i(0, SIZE) pa[i] = (p[i] * SIZE > 1.0) ? 1.0 : p[i] * SIZE; } else if (!strcmp(type, "Hochberg")) { each_i(0, SIZE) pa[i] = i + 1.0; schwartzian(p, pa, UP); } else if (!strcmp(type, "Holm")) { each_i(0, SIZE) pa[i] = SIZE - i; schwartzian(p, pa, DOWN); } else if (!strcmp(type, "Hommel")) { int i, j; int order[SIZE]; int order2[SIZE]; iv1 iv1s[SIZE]; iv2 iv2s[SIZE]; double s[SIZE]; double q[SIZE]; double pa2[SIZE]; int indices[SIZE]; each_i(0, SIZE) { iv1s[i].index = i; iv1s[i].value = p[i]; } qsort(iv1s, SIZE, sizeof(iv1s[0]), compare_iv1); each_i(0, SIZE) order[i] = iv1s[i].index; each_i(0, SIZE) s[i] = p[order[i]]; double min = s[0] * SIZE; each_i(1, SIZE) { double temp = s[i] / (i + 1.0); if (temp < min) min = temp; } each_i(0, SIZE) q[i] = min; each_i(0, SIZE) pa2[i] = min; for (j = SIZE - 1; j >= 2; --j) { each_i(0, SIZE) indices[i] = i; int upper_start = SIZE - j + 1; /* upper indices start index */ int upper_size = j - 1; /* size of upper indices */ int lower_size = SIZE - upper_size; /* size of lower indices */ double qmin = j * s[indices[upper_start]] / 2.0; each_i(1, upper_size) { double temp = s[indices[upper_start + i]] * j / (2.0 + i); if (temp < qmin) qmin = temp; } each_i(0, lower_size) { double temp = s[indices[i]] * j; q[indices[i]] = (temp < qmin) ? temp : qmin; } each_i(0, upper_size) q[indices[upper_start + i]] = q[SIZE - j]; each_i(0, SIZE) if (pa2[i] < q[i]) pa2[i] = q[i]; } each_i(0, SIZE) { iv2s[i].index = i; iv2s[i].value = order[i]; } qsort(iv2s, SIZE, sizeof(iv2s[0]), compare_iv2); each_i(0, SIZE) order2[i] = iv2s[i].index; each_i(0, SIZE) pa[i] = pa2[order2[i]]; } else if (!strcmp(type, "Šidák")) { each_i(0, SIZE) pa[i] = 1.0 - pow(1.0 - p[i], SIZE); } else { printf("\nSorry, do not know how to do '%s' correction.\n", type); printf("Perhaps you want one of these?:\n"); each_i(0, 7) printf(" %s\n", types[i]); exit(1); }
}
void adjusted(const double *p, const char *type) {
int i; double pa[SIZE] = { 0.0 }; if (check(p)) { adjust(p, pa, type); printf("\n%s", type); each_i(0, SIZE) { if (!(i % 5)) printf("\n[%2d] ", i); printf("%1.10f ", pa[i]); } printf("\n"); } else { printf("p-values must be in range 0.0 to 1.0\n"); exit(1); }
}
int check(const double* p) {
int i; each_i(0, SIZE) { if (p[i] < 0.0 || p[i] > 1.0) return 0; } return 1;
}
int main() {
int i; double p_values[SIZE] = { 4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03 }; each_i(0, 8) adjusted(p_values, types[i]); return 0;
}</lang>
- Output:
Same as Kotlin (Version 2) output.
D
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses. <lang D>import std.algorithm; import std.conv; import std.math; import std.stdio; import std.string;
int[] seqLen(int start, int end) {
int[] result; if (start == end) { result.length = end+1; for (int i; i<result.length; i++) { result[i] = i+1; } } else if (start < end) { result.length = end - start + 1; for (int i; i<result.length; i++) { result[i] = start+i; } } else { result.length = start - end + 1; for (int i; i<result.length; i++) { result[i] = start-i; } } return result;
}
int[] order(double[] array, bool decreasing) {
int size = array.length; int[] idx; idx.length = size; double[] baseArr; baseArr.length = size; for (int i; i<size; i++) { baseArr[i] = array[i]; idx[i] = i; } if (!decreasing) { alias comp = (a,b) => baseArr[a] < baseArr[b]; idx.sort!comp; } else { alias comp = (a,b) => baseArr[b] < baseArr[a]; idx.sort!comp; } return idx;
}
double[] cummin(double[] array) {
int size = array.length; if (size < 1) throw new Exception("cummin requires at least one element"); double[] output; output.length = size; auto cumulativeMin = array[0]; foreach (i; 0..size) { if (array[i] < cumulativeMin) cumulativeMin = array[i]; output[i] = cumulativeMin; } return output;
}
double[] cummax(double[] array) {
auto size = array.length; if (size < 1) throw new Exception("cummax requires at least one element"); double[] output; output.length = size; auto cumulativeMax = array[0]; foreach (i; 0..size) { if (array[i] > cumulativeMax) cumulativeMax = array[i]; output[i] = cumulativeMax; } return output;
}
double[] pminx(double[] array, double x) {
auto size = array.length; if (size < 1) throw new Exception("pmin requires at least one element"); double[] result; result.length = size; foreach (i; 0..size) { if (array[i] < x) { result[i] = array[i]; } else { result[i] = x; } } return result;
}
void doubleSay(double[] array) {
writef("[ 1] %e", array[0]); foreach (i; 1..array.length) { writef(" %.10f", array[i]); if ((i+1) % 5 == 0) writef("\n[%2d]", i+1); } writeln;
}
auto toArray(T,U)(U[] array) {
T[] result; result.length = array.length; foreach(i; 0..array.length) { result[i] = to!T(array[i]); } return result;
}
double[] pAdjust(double[] pvalues, string str) {
auto size = pvalues.length; if (size < 1) throw new Exception("pAdjust requires at least one element"); int type = str.toLower.predSwitch!"a==b"( "bh", 0, "fdr", 0, "by", 1, "bonferroni", 2, "hochberg", 3, "holm", 4, "hommel", 5, { throw new Exception(text("'",str,"' doesn't match any accepted FDR types")); }() ); if (type == 2) { // Bonferroni method double[] result; result.length = size; foreach (i; 0..size) { auto b = pvalues[i] * size; if (b >= 1) { result[i] = 1; } else if (0 <= b && b < 1) { result[i] = b; } else { throw new Exception(text(b," is outside [0, 1)")); } } return result; } else if (type == 4) { // Holm method auto o = order(pvalues, false); auto o2Double = toArray!(double,int)(o); double[] cummaxInput; cummaxInput.length = size; foreach (i; 0..size) { cummaxInput[i] = (size-i) * pvalues[o[i]]; } auto ro = order(o2Double, false); auto cummaxOutput = cummax(cummaxInput); auto pmin = pminx(cummaxOutput, 1.0); double[] result; result.length = size; foreach (i; 0..size) { result[i] = pmin[ro[i]]; } return result; } else if (type == 5) { auto indices = seqLen(size, size); auto o = order(pvalues, false); double[] p; p.length = size; foreach (i; 0..size) { p[i] = pvalues[o[i]]; } auto o2Double = toArray!double(o); auto ro = order(o2Double, false); double[] q; q.length = size; double[] pa; pa.length = size; double[] npi; npi.length = size; foreach (i; 0..size) { npi[i] = p[i] * size / indices[i]; } auto min_ = reduce!min(npi); q[] = min_; pa[] = min_; foreach_reverse (j; 2..size) { auto ij = seqLen(1, size - j + 1); foreach (i; 0..size-j+1) { ij[i]--; } auto i2Length = j-1; int[] i2; i2.length = i2Length; foreach(i; 0..i2Length) { i2[i] = size-j+2+i-1; } auto pi2Length = i2Length; double q1 = j*p[i2[0]] / 2.0; foreach (i; 1..pi2Length) { auto temp_q1 = p[i2[i]] * j / (2.0 + i); if (temp_q1 < q1) q1 = temp_q1; } foreach (i; 0..size-j+1) { q[ij[i]] = min(p[ij[i]] * j, q1); } foreach(i; 0..i2Length) { q[i2[i]] = q[size-j]; } foreach(i; 0..size) if (pa[i] < q[i]) pa[i] = q[i]; } foreach (index; 0..size) { q[index] = pa[ro[index]]; } return q; }
double[] ni; ni.length = size; auto o = order(pvalues, true); auto oDouble = toArray!double(o); foreach (index; 0..size) { if (pvalues[index] < 0 || pvalues[index] > 1) { throw new Exception(text("array[", index, "] = ", pvalues[index], " is outside [0, 1]")); } ni[index] = cast(double) size / (size - index); } auto ro = order(oDouble, false); double[] cumminInput; cumminInput.length = size; if (type == 0) { // BH method foreach (index; 0..size) { cumminInput[index] = ni[index] * pvalues[o[index]]; } } else if (type == 1) { // BY method double q = 0; foreach (index; 1..size+1) q += 1.0 / index; foreach (index; 0..size) { cumminInput[index] = q * ni[index] * pvalues[o[index]]; } } else if (type == 3) { // Hochberg method foreach (index; 0..size) { cumminInput[index] = (index + 1) * pvalues[o[index]]; } } auto cumminArray =cummin(cumminInput); auto pmin = pminx(cumminArray, 1.0); double[] result; result.length = size; foreach (i; 0..size) { result[i] = pmin[ro[i]]; } return result;
}
void main() {
double[] pvalues = [ 4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03 ];
double[][] correctAnswers = [ [ // Benjamini-Hochberg 6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02 ], [ // Benjamini & Yekutieli 1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02 ], [ // Bonferroni 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01 ], [ // Hochberg 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 ], [ // Holm 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 ], [ // Hommel 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01 ] ]; auto types = ["bh", "by", "bonferroni", "hochberg", "holm", "hommel"]; foreach (type; 0..types.length) { auto q = pAdjust(pvalues, types[type]); double error = 0.0; foreach (i; 0..pvalues.length) { error += abs(q[i] - correctAnswers[type][i]); } doubleSay(q); writefln("\ntype %d = '%s' has a cumulative error of %g", type, types[type], error); }
}</lang>
- Output:
[ 1] 6.126681e-01 0.8521710465 0.1987205200 0.1891595417 0.3217789286 [ 5] 0.9301450000 0.4870370000 0.9301450000 0.6049730556 0.6826752564 [10] 0.6482628947 0.7253722500 0.5280972727 0.8769925556 0.4705703448 [15] 0.9241867391 0.6049730556 0.7856107317 0.4887525806 0.1136717045 [20] 0.4991890625 0.8769925556 0.9991834000 0.3217789286 0.9301450000 [25] 0.2304957692 0.5832475000 0.0389954722 0.8521710465 0.1476842609 [30] 0.0168363750 0.0025629016 0.0351608437 0.0625018947 0.0036365887 [35] 0.0025629016 0.0294688285 0.0061660636 0.0389954722 0.0026889914 [40] 0.0004502862 0.0000125222 0.0788155476 0.0314261300 0.0048465270 [45] 0.0025629016 0.0048465270 0.0011017083 0.0725203250 0.0220595769 [50] type 0 = 'bh' has a cumulative error of 8.03053e-07 [ 1] 1.000000e+00 1.0000000000 0.8940844244 0.8510676197 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 0.5114323399 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.1754486368 1.0000000000 0.6644618149 [30] 0.0757503082 0.0115310208 0.1581958559 0.2812088585 0.0163617595 [35] 0.0115310208 0.1325863108 0.0277423864 0.1754486368 0.0120983245 [40] 0.0020259303 0.0000563403 0.3546073326 0.1413926119 0.0218055201 [45] 0.0115310208 0.0218055201 0.0049568120 0.3262838334 0.0992505662 [50] type 1 = 'by' has a cumulative error of 3.64072e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.7019185000 1.0000000000 1.0000000000 [30] 0.2020365000 0.0151667450 0.5625735000 1.0000000000 0.0290927100 [35] 0.0153774100 0.4125636000 0.0678267000 0.6803480000 0.0188229400 [40] 0.0009005725 0.0000125222 1.0000000000 0.4713919500 0.0439557650 [45] 0.0108891550 0.0484652700 0.0033051250 1.0000000000 0.2867745000 [50] type 2 = 'bonferroni' has a cumulative error of 6.5e-08 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4632662100 0.9991834000 0.9991834000 [30] 0.1575884700 0.0138396690 0.3938014500 0.7600230400 0.0250197306 [35] 0.0138396690 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125222 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 3 = 'hochberg' has a cumulative error of 2.7375e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.4632662100 1.0000000000 1.0000000000 [30] 0.1575884700 0.0139534054 0.3938014500 0.7600230400 0.0250197306 [35] 0.0139534054 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125222 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 4 = 'holm' has a cumulative error of 2.8095e-07 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9987623800 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9595180000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4351894700 0.9991834000 0.9766522500 [30] 0.1414255500 0.0130434007 0.3530936533 0.6887708800 0.0238560222 [35] 0.0132245726 0.2722919760 0.0542613600 0.4218157600 0.0158112696 [40] 0.0008825610 0.0000125222 0.8743649143 0.3016908480 0.0351646120 [45] 0.0095824564 0.0387722160 0.0031729200 0.8122276400 0.1950066600 [50] type 5 = 'hommel' has a cumulative error of 4.35302e-07
Java
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses. <lang Java>import java.util.Arrays; import java.util.Comparator;
public class PValueCorrection {
private static int[] seqLen(int start, int end) { int[] result; if (start == end) { result = new int[end + 1]; for (int i = 0; i < result.length; ++i) { result[i] = i + 1; } } else if (start < end) { result = new int[end - start + 1]; for (int i = 0; i < result.length; ++i) { result[i] = start + i; } } else { result = new int[start - end + 1]; for (int i = 0; i < result.length; ++i) { result[i] = start - i; } } return result; }
private static int[] order(double[] array, boolean decreasing) { int size = array.length; int[] idx = new int[size]; double[] baseArr = new double[size]; for (int i = 0; i < size; ++i) { baseArr[i] = array[i]; idx[i] = i; }
Comparator<Integer> cmp; if (!decreasing) { cmp = Comparator.comparingDouble(a -> baseArr[a]); } else { cmp = (a, b) -> Double.compare(baseArr[b], baseArr[a]); }
return Arrays.stream(idx) .boxed() .sorted(cmp) .mapToInt(a -> a) .toArray(); }
private static double[] cummin(double[] array) { if (array.length < 1) throw new IllegalArgumentException("cummin requires at least one element"); double[] output = new double[array.length]; double cumulativeMin = array[0]; for (int i = 0; i < array.length; ++i) { if (array[i] < cumulativeMin) cumulativeMin = array[i]; output[i] = cumulativeMin; } return output; }
private static double[] cummax(double[] array) { if (array.length < 1) throw new IllegalArgumentException("cummax requires at least one element"); double[] output = new double[array.length]; double cumulativeMax = array[0]; for (int i = 0; i < array.length; ++i) { if (array[i] > cumulativeMax) cumulativeMax = array[i]; output[i] = cumulativeMax; } return output; }
private static double[] pminx(double[] array, double x) { if (array.length < 1) throw new IllegalArgumentException("pmin requires at least one element"); double[] result = new double[array.length]; for (int i = 0; i < array.length; ++i) { if (array[i] < x) { result[i] = array[i]; } else { result[i] = x; } } return result; }
private static void doubleSay(double[] array) { System.out.printf("[ 1] %e", array[0]); for (int i = 1; i < array.length; ++i) { System.out.printf(" %.10f", array[i]); if ((i + 1) % 5 == 0) System.out.printf("\n[%2d]", i + 1); } System.out.println(); }
private static double[] intToDouble(int[] array) { double[] result = new double[array.length]; for (int i = 0; i < array.length; i++) { result[i] = array[i]; } return result; }
private static double doubleArrayMin(double[] array) { if (array.length < 1) throw new IllegalArgumentException("pAdjust requires at least one element"); return Arrays.stream(array).min().orElse(Double.NaN); }
private static double[] pAdjust(double[] pvalues, String str) { int size = pvalues.length; if (size < 1) throw new IllegalArgumentException("pAdjust requires at least one element"); int type; switch (str.toLowerCase()) { case "bh": case "fdr": type = 0; break; case "by": type = 1; break; case "bonferroni": type = 2; break; case "hochberg": type = 3; break; case "holm": type = 4; break; case "hommel": type = 5; break; default: throw new IllegalArgumentException(str + " doesn't match any accepted FDR types"); }
if (type == 2) { // Bonferroni method double[] result = new double[size]; for (int i = 0; i < size; ++i) { double b = pvalues[i] * size; if (b >= 1) { result[i] = 1; } else if (0 <= b && b < 1) { result[i] = b; } else { throw new RuntimeException("" + b + " is outside [0, 1)"); } } return result; } else if (type == 4) { // Holm method int[] o = order(pvalues, false); double[] o2Double = intToDouble(o); double[] cummaxInput = new double[size]; for (int i = 0; i < size; ++i) { cummaxInput[i] = (size - i) * pvalues[o[i]]; } int[] ro = order(o2Double, false); double[] cummaxOutput = cummax(cummaxInput); double[] pmin = pminx(cummaxOutput, 1.0); double[] result = new double[size]; for (int i = 0; i < size; ++i) { result[i] = pmin[ro[i]]; } return result; } else if (type == 5) { int[] indices = seqLen(size, size); int[] o = order(pvalues, false); double[] p = new double[size]; for (int i = 0; i < size; ++i) { p[i] = pvalues[o[i]]; } double[] o2Double = intToDouble(o); int[] ro = order(o2Double, false); double[] q = new double[size]; double[] pa = new double[size]; double[] npi = new double[size]; for (int i = 0; i < size; ++i) { npi[i] = p[i] * size / indices[i]; } double min = doubleArrayMin(npi); Arrays.fill(q, min); Arrays.fill(pa, min); for (int j = size; j >= 2; --j) { int[] ij = seqLen(1, size - j + 1); for (int i = 0; i < size - j + 1; ++i) { ij[i]--; } int i2Length = j - 1; int[] i2 = new int[i2Length]; for (int i = 0; i < i2Length; ++i) { i2[i] = size - j + 2 + i - 1; } double q1 = j * p[i2[0]] / 2.0; for (int i = 1; i < i2Length; ++i) { double temp_q1 = p[i2[i]] * j / (2.0 + i); if (temp_q1 < q1) q1 = temp_q1; } for (int i = 0; i < size - j + 1; ++i) { q[ij[i]] = Math.min(p[ij[i]] * j, q1); } for (int i = 0; i < i2Length; ++i) { q[i2[i]] = q[size - j]; } for (int i = 0; i < size; ++i) { if (pa[i] < q[i]) { pa[i] = q[i]; } } } for (int i = 0; i < size; ++i) { q[i] = pa[ro[i]]; } return q; }
double[] ni = new double[size]; int[] o = order(pvalues, true); double[] oDouble = intToDouble(o); for (int i = 0; i < size; ++i) { if (pvalues[i] < 0 || pvalues[i] > 1) { throw new RuntimeException("array[" + i + "] = " + pvalues[i] + " is outside [0, 1]"); } ni[i] = (double) size / (size - i); } int[] ro = order(oDouble, false); double[] cumminInput = new double[size]; if (type == 0) { // BH method for (int i = 0; i < size; ++i) { cumminInput[i] = ni[i] * pvalues[o[i]]; } } else if (type == 1) { // BY method double q = 0; for (int i = 1; i < size + 1; ++i) { q += 1.0 / i; } for (int i = 0; i < size; ++i) { cumminInput[i] = q * ni[i] * pvalues[o[i]]; } } else if (type == 3) { // Hochberg method for (int i = 0; i < size; ++i) { cumminInput[i] = (i + 1) * pvalues[o[i]]; } } double[] cumminArray = cummin(cumminInput); double[] pmin = pminx(cumminArray, 1.0); double[] result = new double[size]; for (int i = 0; i < size; ++i) { result[i] = pmin[ro[i]]; } return result; }
public static void main(String[] args) { double[] pvalues = new double[]{ 4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03 };
double[][] correctAnswers = new double[][]{ new double[]{ // Benjamini-Hochberg 6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02 }, new double[]{ // Benjamini & Yekutieli 1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02 }, new double[]{ // Bonferroni 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01 }, new double[]{ // Hochberg 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 }, new double[]{ // Holm 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 }, new double[]{ // Hommel 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01 } };
String[] types = new String[]{"bh", "by", "bonferroni", "hochberg", "holm", "hommel"}; for (int type = 0; type < types.length; ++type) { double[] q = pAdjust(pvalues, types[type]); double error = 0.0; for (int i = 0; i < pvalues.length; ++i) { error += Math.abs(q[i] - correctAnswers[type][i]); } doubleSay(q); System.out.printf("\ntype %d = '%s' has a cumulative error of %g\n", type, types[type], error); } }
}</lang>
- Output:
[ 1] 6.126681e-01 0.8521710465 0.1987205200 0.1891595417 0.3217789286 [ 5] 0.9301450000 0.4870370000 0.9301450000 0.6049730556 0.6826752564 [10] 0.6482628947 0.7253722500 0.5280972727 0.8769925556 0.4705703448 [15] 0.9241867391 0.6049730556 0.7856107317 0.4887525806 0.1136717045 [20] 0.4991890625 0.8769925556 0.9991834000 0.3217789286 0.9301450000 [25] 0.2304957692 0.5832475000 0.0389954722 0.8521710465 0.1476842609 [30] 0.0168363750 0.0025629017 0.0351608438 0.0625018947 0.0036365888 [35] 0.0025629017 0.0294688286 0.0061660636 0.0389954722 0.0026889914 [40] 0.0004502862 0.0000125223 0.0788155476 0.0314261300 0.0048465270 [45] 0.0025629017 0.0048465270 0.0011017083 0.0725203250 0.0220595769 [50] type 0 = 'bh' has a cumulative error of 8.03053e-07 [ 1] 1.000000e+00 1.0000000000 0.8940844244 0.8510676197 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 0.5114323399 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.1754486368 1.0000000000 0.6644618149 [30] 0.0757503083 0.0115310209 0.1581958559 0.2812088585 0.0163617595 [35] 0.0115310209 0.1325863108 0.0277423864 0.1754486368 0.0120983246 [40] 0.0020259303 0.0000563403 0.3546073326 0.1413926119 0.0218055202 [45] 0.0115310209 0.0218055202 0.0049568120 0.3262838334 0.0992505663 [50] type 1 = 'by' has a cumulative error of 3.64072e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.7019185000 1.0000000000 1.0000000000 [30] 0.2020365000 0.0151667450 0.5625735000 1.0000000000 0.0290927100 [35] 0.0153774100 0.4125636000 0.0678267000 0.6803480000 0.0188229400 [40] 0.0009005725 0.0000125223 1.0000000000 0.4713919500 0.0439557650 [45] 0.0108891550 0.0484652700 0.0033051250 1.0000000000 0.2867745000 [50] type 2 = 'bonferroni' has a cumulative error of 6.50000e-08 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4632662100 0.9991834000 0.9991834000 [30] 0.1575884700 0.0138396690 0.3938014500 0.7600230400 0.0250197306 [35] 0.0138396690 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 3 = 'hochberg' has a cumulative error of 2.73750e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.4632662100 1.0000000000 1.0000000000 [30] 0.1575884700 0.0139534054 0.3938014500 0.7600230400 0.0250197306 [35] 0.0139534054 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 4 = 'holm' has a cumulative error of 2.80950e-07 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9987623800 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9595180000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4351894700 0.9991834000 0.9766522500 [30] 0.1414255500 0.0130434007 0.3530936533 0.6887708800 0.0238560222 [35] 0.0132245726 0.2722919760 0.0542613600 0.4218157600 0.0158112696 [40] 0.0008825610 0.0000125223 0.8743649143 0.3016908480 0.0351646120 [45] 0.0095824564 0.0387722160 0.0031729200 0.8122276400 0.1950066600 [50] type 5 = 'hommel' has a cumulative error of 4.35302e-07
Julia
<lang julia>using MultipleTesting using IterTools
p = [4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01,
8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03]
function printpvalues(v)
for chunk in partition(v, 10) println(join((@sprintf("%4.7f", p) for p in chunk), ", ")) end
end
println("Original p-values:") printpvalues(p) for corr in (Bonferroni(), BenjaminiHochberg(), BenjaminiYekutieli(), Holm(), Hochberg(), Hommel())
println("\n", corr) printpvalues(adjust(p, corr))
end</lang>
- Output:
Original p-values: 0.4533744, 0.7296024, 0.0993603, 0.0907966, 0.1801962, 0.8752257, 0.2922222, 0.9115421, 0.4355806, 0.5324867 0.4926798, 0.5802978, 0.3485442, 0.7883130, 0.2729308, 0.8502518, 0.4268138, 0.6442008, 0.3030266, 0.0500155 0.3194810, 0.7892933, 0.9991834, 0.1745691, 0.9037516, 0.1198578, 0.3966083, 0.0140384, 0.7328671, 0.0679348 0.0040407, 0.0003033, 0.0112515, 0.0237507, 0.0005819, 0.0003075, 0.0082513, 0.0013565, 0.0136070, 0.0003765 0.0000180, 0.0000003, 0.0331025, 0.0094278, 0.0008791, 0.0002178, 0.0009693, 0.0000661, 0.0290081, 0.0057355 MultipleTesting.Bonferroni() 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 0.7019185, 1.0000000, 1.0000000 0.2020365, 0.0151667, 0.5625735, 1.0000000, 0.0290927, 0.0153774, 0.4125636, 0.0678267, 0.6803480, 0.0188229 0.0009006, 0.0000125, 1.0000000, 0.4713920, 0.0439558, 0.0108892, 0.0484653, 0.0033051, 1.0000000, 0.2867745 MultipleTesting.BenjaminiHochberg() 0.6126681, 0.8521710, 0.1987205, 0.1891595, 0.3217789, 0.9301450, 0.4870370, 0.9301450, 0.6049731, 0.6826753 0.6482629, 0.7253722, 0.5280973, 0.8769926, 0.4705703, 0.9241867, 0.6049731, 0.7856107, 0.4887526, 0.1136717 0.4991891, 0.8769926, 0.9991834, 0.3217789, 0.9301450, 0.2304958, 0.5832475, 0.0389955, 0.8521710, 0.1476843 0.0168364, 0.0025629, 0.0351608, 0.0625019, 0.0036366, 0.0025629, 0.0294688, 0.0061661, 0.0389955, 0.0026890 0.0004503, 0.0000125, 0.0788155, 0.0314261, 0.0048465, 0.0025629, 0.0048465, 0.0011017, 0.0725203, 0.0220596 MultipleTesting.BenjaminiYekutieli() 1.0000000, 1.0000000, 0.8940844, 0.8510676, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 0.5114323 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 0.1754486, 1.0000000, 0.6644618 0.0757503, 0.0115310, 0.1581959, 0.2812089, 0.0163618, 0.0115310, 0.1325863, 0.0277424, 0.1754486, 0.0120983 0.0020259, 0.0000563, 0.3546073, 0.1413926, 0.0218055, 0.0115310, 0.0218055, 0.0049568, 0.3262838, 0.0992506 MultipleTesting.Holm() 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 1.0000000, 0.4632662, 1.0000000, 1.0000000 0.1575885, 0.0139534, 0.3938014, 0.7600230, 0.0250197, 0.0139534, 0.3052971, 0.0542614, 0.4626366, 0.0165642 0.0008826, 0.0000125, 0.9930759, 0.3394022, 0.0369228, 0.0102358, 0.0397415, 0.0031729, 0.8992520, 0.2179486 MultipleTesting.Hochberg() 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.4632662, 0.9991834, 0.9991834 0.1575885, 0.0138397, 0.3938014, 0.7600230, 0.0250197, 0.0138397, 0.3052971, 0.0542614, 0.4626366, 0.0165642 0.0008826, 0.0000125, 0.9930759, 0.3394022, 0.0369228, 0.0102358, 0.0397415, 0.0031729, 0.8992520, 0.2179486 MultipleTesting.Hommel() 0.9991834, 0.9991834, 0.9991834, 0.9987624, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9595180 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.9991834, 0.4351895, 0.9991834, 0.9766522 0.1414256, 0.0130434, 0.3530937, 0.6887709, 0.0238560, 0.0132246, 0.2722920, 0.0542614, 0.4218158, 0.0158113 0.0008826, 0.0000123, 0.8743649, 0.3016908, 0.0351646, 0.0095825, 0.0387722, 0.0031729, 0.8122276, 0.1950067
Kotlin
Version 1
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses.
<lang scala>// version 1.1.51
import java.util.Arrays
typealias IAE = IllegalArgumentException
fun seqLen(start: Int, end: Int) =
when { start == end -> IntArray(end + 1) { it + 1 } start < end -> IntArray(end - start + 1) { start + it } else -> IntArray(start - end + 1) { start - it } }
var baseArr: DoubleArray? = null
fun compareIncrease(a: Int, b: Int): Int = baseArr!![b].compareTo(baseArr!![a])
fun compareDecrease(a: Int, b: Int): Int = baseArr!![a].compareTo(baseArr!![b])
fun order(array: DoubleArray, decreasing: Boolean): IntArray {
val size = array.size var idx = IntArray(size) { it } baseArr = array.copyOf() if (!decreasing) { idx = Arrays.stream(idx) .boxed() .sorted { a, b -> compareDecrease(a, b) } .mapToInt { it } .toArray() } else { idx = Arrays.stream(idx) .boxed() .sorted { a, b -> compareIncrease(a, b) } .mapToInt { it } .toArray() } baseArr = null return idx
}
fun cummin(array: DoubleArray): DoubleArray {
val size = array.size if (size < 1) throw IAE("cummin requires at least one element") val output = DoubleArray(size) var cumulativeMin = array[0] for (i in 0 until size) { if (array[i] < cumulativeMin) cumulativeMin = array[i] output[i] = cumulativeMin } return output
}
fun cummax(array: DoubleArray): DoubleArray {
val size = array.size if (size < 1) throw IAE("cummax requires at least one element") val output = DoubleArray(size) var cumulativeMax = array[0] for (i in 0 until size) { if (array[i] > cumulativeMax) cumulativeMax = array[i] output[i] = cumulativeMax } return output
}
fun pminx(array: DoubleArray, x: Double): DoubleArray {
val size = array.size if (size < 1) throw IAE("pmin requires at least one element") return DoubleArray(size) { if (array[it] < x) array[it] else x }
}
fun doubleSay(array: DoubleArray) {
print("[ 1] %e".format(array[0])) for (i in 1 until array.size) { print(" %.10f".format(array[i])) if ((i + 1) % 5 == 0) print("\n[%2d]".format(i + 1)) } println()
}
fun intToDouble(array: IntArray) = DoubleArray(array.size) { array[it].toDouble() }
fun doubleArrayMin(array: DoubleArray) =
if (array.size < 1) throw IAE("pAdjust requires at least one element") else array.min()!!
fun pAdjust(pvalues: DoubleArray, str: String): DoubleArray {
val size = pvalues.size if (size < 1) throw IAE("pAdjust requires at least one element") val type = when(str.toLowerCase()) { "bh", "fdr" -> 0 "by" -> 1 "bonferroni" -> 2 "hochberg" -> 3 "holm" -> 4 "hommel" -> 5 else -> throw IAE("'$str' doesn't match any accepted FDR types") } if (type == 2) { // Bonferroni method return DoubleArray(size) { val b = pvalues[it] * size when { b >= 1 -> 1.0 0 <= b && b < 1 -> b else -> throw RuntimeException("$b is outside [0, 1)") } } } else if (type == 4) { // Holm method val o = order(pvalues, false) val o2Double = intToDouble(o) val cummaxInput = DoubleArray(size) { (size - it) * pvalues[o[it]] } val ro = order(o2Double, false) val cummaxOutput = cummax(cummaxInput) val pmin = pminx(cummaxOutput, 1.0) return DoubleArray(size) { pmin[ro[it]] } } else if (type == 5) { // Hommel method val indices = seqLen(size, size) val o = order(pvalues, false) val p = DoubleArray(size) { pvalues[o[it]] } val o2Double = intToDouble(o) val ro = order(o2Double, false) val q = DoubleArray(size) val pa = DoubleArray(size) val npi = DoubleArray(size) { p[it] * size / indices[it] } val min = doubleArrayMin(npi) q.fill(min) pa.fill(min) for (j in size - 1 downTo 2) { val ij = seqLen(1, size - j + 1) for (i in 0 until size - j + 1) ij[i]-- val i2Length = j - 1 val i2 = IntArray(i2Length) { size - j + 2 + it - 1 } val pi2Length = i2Length var q1 = j * p[i2[0]] / 2.0 for (i in 1 until pi2Length) { val temp_q1 = p[i2[i]] * j / (2.0 + i) if(temp_q1 < q1) q1 = temp_q1 } for (i in 0 until size - j + 1) { q[ij[i]] = minOf(p[ij[i]] * j, q1) } for (i in 0 until i2Length) q[i2[i]] = q[size - j] for (i in 0 until size) if (pa[i] < q[i]) pa[i] = q[i] } for (index in 0 until size) q[index] = pa[ro[index]] return q } val ni = DoubleArray(size) val o = order(pvalues, true) val oDouble = intToDouble(o) for (index in 0 until size) { if (pvalues[index] !in 0.0 .. 1.0) { throw RuntimeException("array[$index] = ${pvalues[index]} is outside [0, 1]") } ni[index] = size.toDouble() / (size - index) } val ro = order(oDouble, false) val cumminInput = DoubleArray(size) if (type == 0) { // BH method for (index in 0 until size) { cumminInput[index] = ni[index] * pvalues[o[index]] } } else if (type == 1) { // BY method var q = 0.0 for (index in 1 until size + 1) q += 1.0 / index for (index in 0 until size) { cumminInput[index] = q * ni[index] * pvalues[o[index]] } } else if (type == 3) { // Hochberg method for (index in 0 until size) { cumminInput[index] = (index + 1) * pvalues[o[index]] } } val cumminArray = cummin(cumminInput) val pmin = pminx(cumminArray, 1.0) return DoubleArray(size) { pmin[ro[it]] }
}
fun main(args: Array<String>) {
val pvalues = doubleArrayOf( 4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03 )
val correctAnswers = listOf( doubleArrayOf( // Benjamini-Hochberg 6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02 ), doubleArrayOf( // Benjamini & Yekutieli 1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02 ), doubleArrayOf( // Bonferroni 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01 ), doubleArrayOf( // Hochberg 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 ), doubleArrayOf( // Holm 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01 ), doubleArrayOf( // Hommel 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01 ) ) val types = listOf("bh", "by", "bonferroni", "hochberg", "holm", "hommel") val f = "\ntype %d = '%s' has cumulative error of %g" for (type in 0 until types.size) { val q = pAdjust(pvalues, types[type]) var error = 0.0 for (i in 0 until pvalues.size) { error += Math.abs(q[i] - correctAnswers[type][i]) } doubleSay(q) println(f.format(type, types[type], error)) }
}</lang>
- Output:
[ 1] 6.126681e-01 0.8521710465 0.1987205200 0.1891595417 0.3217789286 [ 5] 0.9301450000 0.4870370000 0.9301450000 0.6049730556 0.6826752564 [10] 0.6482628947 0.7253722500 0.5280972727 0.8769925556 0.4705703448 [15] 0.9241867391 0.6049730556 0.7856107317 0.4887525806 0.1136717045 [20] 0.4991890625 0.8769925556 0.9991834000 0.3217789286 0.9301450000 [25] 0.2304957692 0.5832475000 0.0389954722 0.8521710465 0.1476842609 [30] 0.0168363750 0.0025629017 0.0351608438 0.0625018947 0.0036365888 [35] 0.0025629017 0.0294688286 0.0061660636 0.0389954722 0.0026889914 [40] 0.0004502862 0.0000125223 0.0788155476 0.0314261300 0.0048465270 [45] 0.0025629017 0.0048465270 0.0011017083 0.0725203250 0.0220595769 [50] type 0 = 'bh' has cumulative error of 8.03053e-07 [ 1] 1.000000e+00 1.0000000000 0.8940844244 0.8510676197 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 0.5114323399 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.1754486368 1.0000000000 0.6644618149 [30] 0.0757503083 0.0115310209 0.1581958559 0.2812088585 0.0163617595 [35] 0.0115310209 0.1325863108 0.0277423864 0.1754486368 0.0120983246 [40] 0.0020259303 0.0000563403 0.3546073326 0.1413926119 0.0218055202 [45] 0.0115310209 0.0218055202 0.0049568120 0.3262838334 0.0992505663 [50] type 1 = 'by' has cumulative error of 3.64072e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.7019185000 1.0000000000 1.0000000000 [30] 0.2020365000 0.0151667450 0.5625735000 1.0000000000 0.0290927100 [35] 0.0153774100 0.4125636000 0.0678267000 0.6803480000 0.0188229400 [40] 0.0009005725 0.0000125223 1.0000000000 0.4713919500 0.0439557650 [45] 0.0108891550 0.0484652700 0.0033051250 1.0000000000 0.2867745000 [50] type 2 = 'bonferroni' has cumulative error of 6.50000e-08 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4632662100 0.9991834000 0.9991834000 [30] 0.1575884700 0.0138396690 0.3938014500 0.7600230400 0.0250197306 [35] 0.0138396690 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 3 = 'hochberg' has cumulative error of 2.73750e-07 [ 1] 1.000000e+00 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.4632662100 1.0000000000 1.0000000000 [30] 0.1575884700 0.0139534054 0.3938014500 0.7600230400 0.0250197306 [35] 0.0139534054 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825610 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 [50] type 4 = 'holm' has cumulative error of 2.80950e-07 [ 1] 9.991834e-01 0.9991834000 0.9991834000 0.9987623800 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9595180000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4351894700 0.9991834000 0.9766522500 [30] 0.1414255500 0.0130434007 0.3530936533 0.6887708800 0.0238560222 [35] 0.0132245726 0.2722919760 0.0542613600 0.4218157600 0.0158112696 [40] 0.0008825610 0.0000125223 0.8743649143 0.3016908480 0.0351646120 [45] 0.0095824564 0.0387722160 0.0031729200 0.8122276400 0.1950066600 [50] type 5 = 'hommel' has cumulative error of 4.35302e-07
Version 2
To avoid licensing issues, this version follows the approach of the Perl 6 entry of which it is a partial translation. However, the correction routines themselves have been coded independently, common code factored out into separate functions (analogous to Perl 6) and (apart from the Šidák method) agree with the Perl 6 results. <lang scala>// version 1.2.21
typealias DList = List<Double> typealias DToD = (Double) -> Double
enum class Direction { UP, DOWN }
// test also for 'Unknown' correction type val types = listOf(
"Benjamini-Hochberg", "Benjamini-Yekutieli", "Bonferroni", "Hochberg", "Holm", "Hommel", "Šidák", "Unknown"
)
fun adjusted(p: DList, type: String) = "\n$type\n${pFormat(adjust(check(p), type))}"
fun pFormat(p: DList, cols: Int = 5): String {
var i = -cols val fmt = "%1.10f" return p.chunked(cols).map { chunk -> i += cols "[%2d] %s".format(i, chunk.map { fmt.format(it) }.joinToString(" ")) }.joinToString("\n")
}
fun check(p: DList): DList {
require(p.size > 0 && p.min()!! >= 0.0 && p.max()!! <= 1.0) { "p-values must be in range 0.0 to 1.0" } return p
}
fun ratchet(p: DList, dir: Direction): DList {
val pp = p.toMutableList() var m = pp[0] if (dir == Direction.UP) { for (i in 1 until pp.size) { if (pp[i] > m) pp[i] = m m = pp[i] } } else { for (i in 1 until pp.size) { if (pp[i] < m) pp[i] = m m = pp[i] } } return pp.map { if (it < 1.0) it else 1.0 }
}
fun schwartzian(p: DList, mult: DList, dir: Direction): DList {
val size = p.size val order = if (dir == Direction.UP) p.withIndex().sortedByDescending { it.value }.map { it.index } else p.withIndex().sortedBy { it.value }.map { it.index } var pa = List(size) { mult[it] * p[order[it]] } pa = ratchet(pa, dir) val order2 = order.withIndex().sortedBy{ it.value }.map { it.index } return List(size) { pa[order2[it]] }
}
fun adjust(p: DList, type: String): DList {
val size = p.size require(size > 0) when (type) { "Benjamini-Hochberg" -> { val mult = List(size) { size.toDouble() / (size - it) } return schwartzian(p, mult, Direction.UP) }
"Benjamini-Yekutieli" -> { val q = (1..size).sumByDouble { 1.0 / it } val mult = List(size) { q * size / (size - it) } return schwartzian(p, mult, Direction.UP) }
"Bonferroni" -> { return p.map { minOf(it * size, 1.0) } }
"Hochberg" -> { val mult = List(size) { (it + 1).toDouble() } return schwartzian(p, mult, Direction.UP) }
"Holm" -> { val mult = List(size) { (size - it).toDouble() } return schwartzian(p, mult, Direction.DOWN) }
"Hommel" -> { val order = p.withIndex().sortedBy { it.value }.map { it.index } val s = List(size) { p[order[it]] } val min = List(size){ s[it] * size / ( it + 1) }.min()!! val q = MutableList(size) { min } val pa = MutableList(size) { min } for (j in size - 1 downTo 2) { val lower = IntArray(size - j + 1) { it } // lower indices val upper = IntArray(j - 1) { size - j + 1 + it } // upper indices var qmin = j * s[upper[0]] / 2.0 for (i in 1 until upper.size) { val temp = s[upper[i]] * j / (2.0 + i) if (temp < qmin) qmin = temp } for (i in 0 until lower.size) { q[lower[i]] = minOf(s[lower[i]] * j, qmin) } for (i in 0 until upper.size) q[upper[i]] = q[size - j] for (i in 0 until size) if (pa[i] < q[i]) pa[i] = q[i] } val order2 = order.withIndex().sortedBy{ it.value }.map { it.index } return List(size) { pa[order2[it]] } }
"Šidák" -> { val m = size.toDouble() return p.map { 1.0 - Math.pow(1.0 - it, m) } }
else -> { println( "\nSorry, do not know how to do '$type' correction.\n" + "Perhaps you want one of these?:\n" + types.dropLast(1).map { " $it" }.joinToString("\n") ) System.exit(1) } } return p
}
fun main(args: Array<String>) {
val pValues = listOf( 4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03 )
types.forEach { println(adjusted(pValues, it)) }
}</lang>
- Output:
Same as Perl 6 entry except:
.... Šidák [ 0] 1.0000000000 1.0000000000 0.9946598274 0.9914285749 0.9999515274 [ 5] 1.0000000000 0.9999999688 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 0.9999999995 1.0000000000 0.9999998801 [15] 1.0000000000 1.0000000000 1.0000000000 0.9999999855 0.9231179729 [20] 0.9999999956 1.0000000000 1.0000000000 0.9999317605 1.0000000000 [25] 0.9983109511 1.0000000000 0.5068253940 1.0000000000 0.9703301333 [30] 0.1832692440 0.0150545753 0.4320729669 0.6993672225 0.0286818157 [35] 0.0152621104 0.3391808707 0.0656206307 0.4959194266 0.0186503726 [40] 0.0009001752 0.0000125222 0.8142104886 0.3772612062 0.0430222116 [45] 0.0108312558 0.0473319661 0.0032997780 0.7705015898 0.2499384839 Sorry, do not know how to do 'Unknown' correction. Perhaps you want one of these?: Benjamini-Hochberg Benjamini-Yekutieli Bonferroni Hochberg Holm Hommel Šidák
Perl
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses. <lang perl>#!/usr/bin/env perl
use strict; use warnings; use Cwd 'getcwd'; my $TOP_DIRECTORY = getcwd();
sub log_error_and_die {
my $error = shift;
edirect=1#comment345753_182010
my $fail_filename = "$TOP_DIRECTORY/$0.FAIL"; open my $fh, '>', $fail_filename or die "Can't write $fail_filename: $!"; print $fh $error;
die $error;
}
local $SIG{__WARN__} = sub {
my $message = shift; log_error_and_die( sprintf( '%s @ %s', $message, getcwd() ) );
}; sub pmin { my $array_ref = shift; my $x = 1; unless ((ref $array_ref) =~ m/ARRAY/) { print "cummin requires an array.\n"; die; } my @pmin_array; my $n = scalar @$array_ref; for (my $index = 0; $index < $n; $index++) { if (@$array_ref[$index] < $x) { $pmin_array[$index] = @$array_ref[$index]; } else { $pmin_array[$index] = $x; } } return @pmin_array; }
sub cummin { my $array_ref = shift; unless ((ref $array_ref) =~ m/ARRAY/) { print "cummin requires an array.\n"; die; } my @cummin; my $cumulative_min = @$array_ref[0]; foreach my $p (@$array_ref) { if ($p < $cumulative_min) { $cumulative_min = $p; } push @cummin, $cumulative_min; } return @cummin; }
sub cummax { my $array_ref = shift; unless ((ref $array_ref) =~ m/ARRAY/) { print "cummin requires an array.\n"; die; } my @cummax; my $cumulative_max = @$array_ref[0]; foreach my $p (@$array_ref) { if ($p > $cumulative_max) { $cumulative_max = $p; } push @cummax, $cumulative_max; } return @cummax; }
sub order {#made to match R's "order" my $array_ref = shift; my $decreasing = 'false'; if (defined $_[0]) { my $option = shift; if ($option =~ m/true/i) { $decreasing = 'true'; } elsif ($option =~ m/false/i) { #do nothing, it's already set to false } else { print "2nd option should only be case-insensitive 'true' or 'false'"; die; } } unless ((ref $array_ref) =~ m/ARRAY/) { print "You should have entered an array.\n"; die; } my @array; my $max_index = scalar @$array_ref-1; if ($decreasing eq 'false') { @array = sort { @$array_ref[$a] <=> @$array_ref[$b] } 0..$max_index; } elsif ($decreasing eq 'true') { @array = sort { @$array_ref[$b] <=> @$array_ref[$a] } 0..$max_index; } }
sub min { my $array_ref = shift; unless ((ref $array_ref) =~ m/ARRAY/) { print "min requires an array.\n"; die; } my $min = @$array_ref[0]; foreach my $e (@$array_ref) { if ($e < $min) { $min = $e; } } return $min; }
sub min2 { my $n1 = shift; my $n2 = shift; if ($n1 < $n2) { return $n1; } else { return $n2; } }
sub p_adjust { my $pvalues_ref = shift; unless ((ref $pvalues_ref) =~ m/ARRAY/) { print "p_adjust requires an array.\n"; die; } my $method; if (defined $_[0]) { $method = shift; } else { $method = 'Holm'; } my %methods = ( 'bh' => 1, 'fdr' => 1, 'by' => 1, 'holm' => 1, 'hommel' => 1, 'bonferroni' => 1, 'hochberg' => 1 ); my $method_found = 'no'; foreach my $key (keys %methods) { if ((uc $method) eq (uc $key)) { $method = $key; $method_found = 'yes'; last; } } if ($method_found eq 'no') { if ($method =~ m/benjamini-?\s*hochberg/i) { $method = 'bh'; $method_found = 'yes'; } elsif ($method =~ m/benjamini-?\s*yekutieli/i) { $method = 'by'; $method_found = 'yes'; } } if ($method_found eq 'no') { print "No method could be determined from $method.\n"; die; } my $lp = scalar @$pvalues_ref; my $n = $lp; my @qvalues; if ($method eq 'hochberg') { my @o = order($pvalues_ref, 'TRUE'); my @cummin_input; for (my $index = 0; $index < $n; $index++) { $cummin_input[$index] = ($index+1)* @$pvalues_ref[$o[$index]];#PVALUES[$o[$index]] is p[o] } my @cummin = cummin(\@cummin_input); undef @cummin_input; my @pmin = pmin(\@cummin); undef @cummin; my @ro = order(\@o); undef @o; @qvalues = @pmin[@ro]; } elsif ($method eq 'bh') { my @o = order($pvalues_ref, 'TRUE'); my @cummin_input; for (my $index = 0; $index < $n; $index++) { $cummin_input[$index] = ($n/($n-$index))* @$pvalues_ref[$o[$index]];#PVALUES[$o[$index]] is p[o] } my @ro = order(\@o); undef @o; my @cummin = cummin(\@cummin_input); undef @cummin_input; my @pmin = pmin(\@cummin); undef @cummin; @qvalues = @pmin[@ro]; } elsif ($method eq 'by') { my $q = 0.0; my @o = order($pvalues_ref, 'TRUE'); my @ro = order(\@o); for (my $index = 1; $index < ($n+1); $index++) { $q += 1.0 / $index; } my @cummin_input; for (my $index = 0; $index < $n; $index++) { $cummin_input[$index] = $q * ($n/($n-$index)) * @$pvalues_ref[$o[$index]];#PVALUES[$o[$index]] is p[o] } my @cummin = cummin(\@cummin_input); undef @cummin_input; my @pmin = pmin(\@cummin); undef @cummin; @qvalues = @pmin[@ro]; } elsif ($method eq 'bonferroni') { for (my $index = 0; $index < $n; $index++) { my $q = @$pvalues_ref[$index]*$n; if ((0 <= $q) && ($q < 1)) { $qvalues[$index] = $q; } elsif ($q >= 1) { $qvalues[$index] = 1.0; } else { print "Failed to get Bonferroni adjusted p."; die; } } } elsif ($method eq 'holm') { my @o = order($pvalues_ref); my @cummax_input; for (my $index = 0; $index < $n; $index++) { $cummax_input[$index] = ($n - $index) * @$pvalues_ref[$o[$index]]; } my @ro = order(\@o); undef @o; my @cummax = cummax(\@cummax_input); undef @cummax_input; my @pmin = pmin(\@cummax); undef @cummax; @qvalues = @pmin[@ro]; } elsif ($method eq 'hommel') { my @i = 1..$n; my @o = order($pvalues_ref); my @p = @$pvalues_ref[@o]; my @ro = order(\@o); undef @o; my @pa; my @q; my $min = $n*$p[0]; for (my $index = 0; $index < $n; $index++) { my $temp = $n*$p[$index] / ($index + 1); if ($temp < $min) { $min = $temp; } } for (my $index = 0; $index < $n; $index++) { $pa[$index] = $min;#q <- pa <- rep.int(min(n * p/i), n) $q[$index] = $min;#q <- pa <- rep.int(min(n * p/i), n) } for (my $j = ($n-1); $j >= 2; $j--) {
- printf("j = %zu\n", j);
my @ij = 1..($n - $j + 1);#ij <- seq_len(n - j + 1) for (my $i = 0; $i < $n - $j + 1; $i++) { $ij[$i]--;#R's indices are 1-based, C's are 0-based } my $I2_LENGTH = $j - 1; my @i2; for (my $i = 0; $i < $I2_LENGTH; $i++) { $i2[$i] = $n-$j+2+$i-1;
- R's indices are 1-based, C's are 0-based, I added the -1
}
my $q1 = $j * $p[$i2[0]] / 2.0; for (my $i = 1; $i < $I2_LENGTH; $i++) {#loop through 2:j my $TEMP_Q1 = $j * $p[$i2[$i]] / (2 + $i); if ($TEMP_Q1 < $q1) { $q1 = $TEMP_Q1; } }
for (my $i = 0; $i < ($n - $j + 1); $i++) {#q[ij] <- pmin(j * p[ij], q1) $q[$ij[$i]] = min2( $j*$p[$ij[$i]], $q1); }
for (my $i = 0; $i < $I2_LENGTH; $i++) {#q[i2] <- q[n - j + 1] $q[$i2[$i]] = $q[$n - $j];#subtract 1 because of starting index difference }
for (my $i = 0; $i < $n; $i++) {#pa <- pmax(pa, q) if ($pa[$i] < $q[$i]) { $pa[$i] = $q[$i]; } }
- printf("j = %zu, pa = \n", j);
- double_say(pa, N);
}#end j loop @qvalues = @pa[@ro]; } else { print "$method doesn't fit my types.\n"; die; } return @qvalues; } my @pvalues = (4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03);
my %correct_answers = ( 'Benjamini-Hochberg' => [6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02], 'Benjamini-Yekutieli' => [1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02], 'Bonferroni' => [1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01],
'Hochberg' => [9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01], 'Holm' => [1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01],
'Hommel' => [9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01]);
foreach my $method ('Hochberg','Benjamini-Hochberg','Benjamini-Yekutieli', 'Bonferroni', 'Holm', 'Hommel') {
print "$method\n";
my @qvalues = p_adjust(\@pvalues, $method);
my $error = 0.0;
foreach my $q (0..$#qvalues) {
$error += abs($qvalues[$q] - $correct_answers{$method}[$q]);
}
printf("type $method has cumulative error of %g.\n", $error);
}
</lang>
- Output:
Hochberg type Hochberg has cumulative error of 2.7375e-07. Benjamini-Hochberg type Benjamini-Hochberg has cumulative error of 8.03053e-07. Benjamini-Yekutieli type Benjamini-Yekutieli has cumulative error of 3.64072e-07. Bonferroni type Bonferroni has cumulative error of 6.5e-08. Holm type Holm has cumulative error of 2.8095e-07. Hommel type Hommel has cumulative error of 4.35302e-07.
Perl 6
<lang perl6>########################### Helper subs ###########################
sub adjusted (@p, $type) { "\n$type\n" ~ format adjust( check(@p), $type ) }
sub format ( @p, $cols = 5 ) {
my $i = -$cols; my $fmt = "%1.10f"; join "\n", @p.rotor($cols, :partial).map: { sprintf "[%2d] { join ' ', $fmt xx $_ }", $i+=$cols, $_ };
}
sub check ( @p ) { die 'p-values must be in range 0.0 to 1.0' if @p.min < 0 or 1 < @p.max; @p }
multi ratchet ( 'up', @p ) { my $m; @p[$_] min= $m, $m = @p[$_] for ^@p; @p }
multi ratchet ( 'dn', @p ) { my $m; @p[$_] max= $m, $m = @p[$_] for ^@p .reverse; @p }
sub schwartzian ( @p, &transform, :$ratchet ) {
my @pa = @p.map( {[$_, $++]} ).sort( -*.[0] ).map: { [transform(.[0]), .[1]] }; @pa[*;0] = ratchet($ratchet, @pa»[0]); @pa.sort( *.[1] )»[0]
}
- The various p-value correction routines #############
multi adjust( @p, 'Benjamini-Hochberg' ) {
@p.&schwartzian: * * @p / (@p - $++) min 1, :ratchet('up')
}
multi adjust( @p, 'Benjamini-Yekutieli' ) {
my \r = ^@p .map( { 1 / ++$ } ).sum; @p.&schwartzian: * * r * @p / (@p - $++) min 1, :ratchet('up')
}
multi adjust( @p, 'Hochberg' ) {
my \m = @p.max; @p.&schwartzian: * * ++$ min m, :ratchet('up')
}
multi adjust( @p, 'Holm' ) {
@p.&schwartzian: * * ++$ min 1, :ratchet('dn')
}
multi adjust( @p, 'Šidák' ) {
@p.&schwartzian: 1 - (1 - *) ** ++$, :ratchet('dn')
}
multi adjust( @p, 'Bonferroni' ) {
@p.map: * * @p min 1
}
- Hommel correction can't be easily reduced to a one pass transform
multi adjust( @p, 'Hommel' ) {
my @s = @p.map( {[$_, $++]} ).sort: *.[0] ; # sorted my \z = +@p; # array si(z)e my @pa = @s»[0].map( * * z / ++$ ).min xx z; # p adjusted my @q; # scratch array for (1 ..^ z).reverse -> $i { my @L = 0 .. z - $i; # lower indices my @U = z - $i ^..^ z; # upper indices my $q = @s[@U]»[0].map( { $_ * $i / (2 + $++) } ).min; @q[@L] = @s[@L]»[0].map: { [min] $_ * $i, $q, @s[*-1][0] }; @pa = ^z .map: { [max] @pa[$_], @q[$_] } } @pa[@s[*;1].map( {[$_, $++]} ).sort( *.[0] )»[1]]
}
multi adjust ( @p, $unknown ) {
note "\nSorry, do not know how to do $unknown correction.\n" ~ "Perhaps you want one of these?:\n" ~ <Benjamini-Hochberg Benjamini-Yekutieli Bonferroni Hochberg Holm Hommel Šidák>.join("\n"); exit
}
- The task ###########################
my @p-values =
4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03
for < Benjamini-Hochberg Benjamini-Yekutieli Bonferroni Hochberg Holm Hommel Šidák > {
say adjusted @p-values, $_
}</lang>
- Output:
Benjamini-Hochberg [ 0] 0.6126681081 0.8521710465 0.1987205200 0.1891595417 0.3217789286 [ 5] 0.9301450000 0.4870370000 0.9301450000 0.6049730556 0.6826752564 [10] 0.6482628947 0.7253722500 0.5280972727 0.8769925556 0.4705703448 [15] 0.9241867391 0.6049730556 0.7856107317 0.4887525806 0.1136717045 [20] 0.4991890625 0.8769925556 0.9991834000 0.3217789286 0.9301450000 [25] 0.2304957692 0.5832475000 0.0389954722 0.8521710465 0.1476842609 [30] 0.0168363750 0.0025629017 0.0351608438 0.0625018947 0.0036365888 [35] 0.0025629017 0.0294688286 0.0061660636 0.0389954722 0.0026889914 [40] 0.0004502863 0.0000125223 0.0788155476 0.0314261300 0.0048465270 [45] 0.0025629017 0.0048465270 0.0011017083 0.0725203250 0.0220595769 Benjamini-Yekutieli [ 0] 1.0000000000 1.0000000000 0.8940844244 0.8510676197 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 0.5114323399 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.1754486368 1.0000000000 0.6644618149 [30] 0.0757503083 0.0115310209 0.1581958559 0.2812088585 0.0163617595 [35] 0.0115310209 0.1325863108 0.0277423864 0.1754486368 0.0120983246 [40] 0.0020259303 0.0000563403 0.3546073326 0.1413926119 0.0218055202 [45] 0.0115310209 0.0218055202 0.0049568120 0.3262838334 0.0992505663 Bonferroni [ 0] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.7019185000 1.0000000000 1.0000000000 [30] 0.2020365000 0.0151667450 0.5625735000 1.0000000000 0.0290927100 [35] 0.0153774100 0.4125636000 0.0678267000 0.6803480000 0.0188229400 [40] 0.0009005725 0.0000125223 1.0000000000 0.4713919500 0.0439557650 [45] 0.0108891550 0.0484652700 0.0033051250 1.0000000000 0.2867745000 Hochberg [ 0] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4632662100 0.9991834000 0.9991834000 [30] 0.1575884700 0.0138396690 0.3938014500 0.7600230400 0.0250197306 [35] 0.0138396690 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825611 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 Holm [ 0] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [ 5] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [10] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [15] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [20] 1.0000000000 1.0000000000 1.0000000000 1.0000000000 1.0000000000 [25] 1.0000000000 1.0000000000 0.4632662100 1.0000000000 1.0000000000 [30] 0.1575884700 0.0139534054 0.3938014500 0.7600230400 0.0250197306 [35] 0.0139534054 0.3052970640 0.0542613600 0.4626366400 0.0165641872 [40] 0.0008825611 0.0000125223 0.9930759000 0.3394022040 0.0369228426 [45] 0.0102358057 0.0397415214 0.0031729200 0.8992520300 0.2179486200 Hommel [ 0] 0.9991834000 0.9991834000 0.9991834000 0.9987623800 0.9991834000 [ 5] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [10] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [15] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9595180000 [20] 0.9991834000 0.9991834000 0.9991834000 0.9991834000 0.9991834000 [25] 0.9991834000 0.9991834000 0.4351894700 0.9991834000 0.9766522500 [30] 0.1414255500 0.0130434007 0.3530936533 0.6887708800 0.0238560222 [35] 0.0132245726 0.2722919760 0.0542613600 0.4218157600 0.0158112696 [40] 0.0008825611 0.0000125223 0.8743649143 0.3016908480 0.0351646120 [45] 0.0095824564 0.0387722160 0.0031729200 0.8122276400 0.1950066600 Šidák [ 0] 0.9998642526 0.9999922727 0.9341844137 0.9234670175 0.9899922294 [ 5] 0.9999922727 0.9992955735 0.9999922727 0.9998642526 0.9998909746 [10] 0.9998642526 0.9999288207 0.9995533892 0.9999922727 0.9990991210 [15] 0.9999922727 0.9998642526 0.9999674876 0.9992955735 0.7741716825 [20] 0.9993332472 0.9999922727 0.9999922727 0.9899922294 0.9999922727 [25] 0.9589019598 0.9998137104 0.3728369461 0.9999922727 0.8605248833 [30] 0.1460714182 0.0138585952 0.3270159382 0.5366136349 0.0247164330 [35] 0.0138585952 0.2640282766 0.0528503728 0.3723753774 0.0164308228 [40] 0.0008821796 0.0000125222 0.6357389664 0.2889497995 0.0362651575 [45] 0.0101847015 0.0389807074 0.0031679962 0.5985019850 0.1963376344
Python
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses. <lang python>from __future__ import division import sys
def pminf(array):
x = 1 pmin_list = [] N = len(array) for index in range(N): if array[index] < x: pmin_list.insert(index, array[index]) else: pmin_list.insert(index, x) return pmin_list
- end function
def cumminf(array):
cummin = [] cumulative_min = array[0] for p in array: if p < cumulative_min: cumulative_min = p cummin.append(cumulative_min) return cummin
- end
def cummaxf(array):
cummax = [] cumulative_max = array[0] for e in array: if e > cumulative_max: cumulative_max = e cummax.append(cumulative_max) return cummax
- end
def order(*args):
if len(args) > 1: if args[1].lower() == 'false':# if ($string1 eq $string2) { return sorted(range(len(args[0])), key = lambda k: args[0][k]) elif list(args[1].lower()) == list('true'): return sorted(range(len(args[0])), key = lambda k: args[0][k], reverse = True) else: print "%s isn't a recognized parameter" % args[1] sys.exit() elif len(args) == 1: return sorted(range(len(args[0])), key = lambda k: args[0][k])
- end
def p_adjust(*args):
method = "bh" pvalues = args[0] if len(args) > 1: methods = {"bh", "fdr", "by", "holm", "hommel", "bonferroni", "hochberg"} metharg = arg[1].lower() if metharg in methods: method = metharg lp = len(pvalues) n = lp qvalues = []
if method == 'hochberg':#already all lower case o = order(pvalues, 'TRUE') cummin_input = [] for index in range(n): cummin_input.insert(index, (index+1)*pvalues[o[index]]) cummin = cumminf(cummin_input) pmin = pminf(cummin) ro = order(o) qvalues = [pmin[i] for i in ro] elif method == 'bh': o = order(pvalues, 'TRUE') cummin_input = [] for index in range(n): cummin_input.insert(index, (n/(n-index))* pvalues[o[index]]) ro = order(o) cummin = cumminf(cummin_input) pmin = pminf(cummin) qvalues = [pmin[i] for i in ro] elif method == 'by': q = 0.0 o = order(pvalues, 'TRUE') ro = order(o) for index in range(1, n+1): q += 1.0 / index; cummin_input = [] for index in range(n): cummin_input.insert(index, q * (n/(n-index)) * pvalues[o[index]]) cummin = cumminf(cummin_input) pmin = pminf(cummin) qvalues = [pmin[i] for i in ro] elif method == 'bonferroni': for index in range(n): q = pvalues[index] * n if (0 <= q) and (q < 1): qvalues.insert(index, q) elif q >= 1: qvalues.insert(index, 1) else: print '%g won\'t give a Bonferroni adjusted p' % q sys.exit() elif method == 'holm': o = order(pvalues) cummax_input = [] for index in range(n): cummax_input.insert(index, (n - index) * pvalues[o[index]]) ro = order(o) cummax = cummaxf(cummax_input) pmin = pminf(cummax) qvalues = [pmin[i] for i in ro] elif method == 'hommel': i = range(1,n+1) o = order(pvalues) p = [pvalues[index] for index in o] ro = order(o) pa = [] q = [] smin = n*p[0] for index in range(n): temp = n*p[index] / (index + 1) if temp < smin: smin = temp for index in range(n): pa.insert(index, smin) q.insert(index, smin) for j in range(n-1,1,-1): ij = range(1,n-j+2) for x in range(len(ij)): ij[x] -= 1 I2_LENGTH = j - 1 i2 = [] for index in range(I2_LENGTH+1): i2.insert(index, n - j + 2 + index - 1) q1 = j * p[i2[0]] / 2.0 for index in range(1,I2_LENGTH): TEMP_Q1 = j * p[i2[index]] / (2.0 + index) if TEMP_Q1 < q1: q1 = TEMP_Q1 for index in range(n - j + 1): q[ij[index]] = min(j * p[ij[index]], q1) for index in range(I2_LENGTH): q[i2[index]] = q[n-j] for index in range(n): if pa[index] < q[index]: pa[index] = q[index] qvalues = [pa[index] for index in ro] else: print "method %s isn't defined." % method sys.exit() return qvalues
pvalues = [4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03]
correct_answers = {}
correct_answers['bh'] = [6.126681e-01, 8.521710e-01, 1.987205e-01, 1.891595e-01, 3.217789e-01, 9.301450e-01, 4.870370e-01, 9.301450e-01, 6.049731e-01, 6.826753e-01, 6.482629e-01, 7.253722e-01, 5.280973e-01, 8.769926e-01, 4.705703e-01, 9.241867e-01, 6.049731e-01, 7.856107e-01, 4.887526e-01, 1.136717e-01, 4.991891e-01, 8.769926e-01, 9.991834e-01, 3.217789e-01, 9.301450e-01, 2.304958e-01, 5.832475e-01, 3.899547e-02, 8.521710e-01, 1.476843e-01, 1.683638e-02, 2.562902e-03, 3.516084e-02, 6.250189e-02, 3.636589e-03, 2.562902e-03, 2.946883e-02, 6.166064e-03, 3.899547e-02, 2.688991e-03, 4.502862e-04, 1.252228e-05, 7.881555e-02, 3.142613e-02, 4.846527e-03, 2.562902e-03, 4.846527e-03, 1.101708e-03, 7.252032e-02, 2.205958e-02]
correct_answers['by'] = [1.000000e+00, 1.000000e+00, 8.940844e-01, 8.510676e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 5.114323e-01, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.754486e-01, 1.000000e+00, 6.644618e-01, 7.575031e-02, 1.153102e-02, 1.581959e-01, 2.812089e-01, 1.636176e-02, 1.153102e-02, 1.325863e-01, 2.774239e-02, 1.754486e-01, 1.209832e-02, 2.025930e-03, 5.634031e-05, 3.546073e-01, 1.413926e-01, 2.180552e-02, 1.153102e-02, 2.180552e-02, 4.956812e-03, 3.262838e-01, 9.925057e-02]
correct_answers['bonferroni'] = [1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 7.019185e-01, 1.000000e+00, 1.000000e+00, 2.020365e-01, 1.516674e-02, 5.625735e-01, 1.000000e+00, 2.909271e-02, 1.537741e-02, 4.125636e-01, 6.782670e-02, 6.803480e-01, 1.882294e-02, 9.005725e-04, 1.252228e-05, 1.000000e+00, 4.713920e-01, 4.395577e-02, 1.088915e-02, 4.846527e-02, 3.305125e-03, 1.000000e+00, 2.867745e-01]
correct_answers['hochberg'] = [9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.632662e-01, 9.991834e-01, 9.991834e-01, 1.575885e-01, 1.383967e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.383967e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01]
correct_answers['holm'] = [1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 1.000000e+00, 4.632662e-01, 1.000000e+00, 1.000000e+00, 1.575885e-01, 1.395341e-02, 3.938014e-01, 7.600230e-01, 2.501973e-02, 1.395341e-02, 3.052971e-01, 5.426136e-02, 4.626366e-01, 1.656419e-02, 8.825610e-04, 1.252228e-05, 9.930759e-01, 3.394022e-01, 3.692284e-02, 1.023581e-02, 3.974152e-02, 3.172920e-03, 8.992520e-01, 2.179486e-01]
correct_answers['hommel'] = [9.991834e-01, 9.991834e-01, 9.991834e-01, 9.987624e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.595180e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 9.991834e-01, 4.351895e-01, 9.991834e-01, 9.766522e-01, 1.414256e-01, 1.304340e-02, 3.530937e-01, 6.887709e-01, 2.385602e-02, 1.322457e-02, 2.722920e-01, 5.426136e-02, 4.218158e-01, 1.581127e-02, 8.825610e-04, 1.252228e-05, 8.743649e-01, 3.016908e-01, 3.516461e-02, 9.582456e-03, 3.877222e-02, 3.172920e-03, 8.122276e-01, 1.950067e-01]
for key in correct_answers.keys():
error = 0.0 q = p_adjust(pvalues, key) for i in range(len(q)): error += abs(q[i] - correct_answers[key][i]) print '%s error = %g' % (key.upper(), error)
</lang>
- Output:
BONFERRONI error = 6.5e-08 BH error = 8.03053e-07 HOLM error = 2.8095e-07 HOMMEL error = 4.35302e-07 HOCHBERG error = 2.7375e-07 BY error = 3.64072e-07
R
The p.adjust function is built-in, see R manual.
<lang R>p <- c(4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03)
p.adjust(p, method = 'BH') print("Benjamini-Hochberg") writeLines("\n")
p.adjust(p, method = 'BY') print("Benjamini & Yekutieli") writeLines("\n")
p.adjust(p, method = 'bonferroni') print("Bonferroni") writeLines("\n")
p.adjust(p, method = 'hochberg') print("Hochberg") writeLines("\n");
p.adjust(p, method = 'hommel') writeLines("Hommel\n")</lang>
- Output:
[1] 6.126681e-01 8.521710e-01 1.987205e-01 1.891595e-01 3.217789e-01 [6] 9.301450e-01 4.870370e-01 9.301450e-01 6.049731e-01 6.826753e-01 [11] 6.482629e-01 7.253722e-01 5.280973e-01 8.769926e-01 4.705703e-01 [16] 9.241867e-01 6.049731e-01 7.856107e-01 4.887526e-01 1.136717e-01 [21] 4.991891e-01 8.769926e-01 9.991834e-01 3.217789e-01 9.301450e-01 [26] 2.304958e-01 5.832475e-01 3.899547e-02 8.521710e-01 1.476843e-01 [31] 1.683638e-02 2.562902e-03 3.516084e-02 6.250189e-02 3.636589e-03 [36] 2.562902e-03 2.946883e-02 6.166064e-03 3.899547e-02 2.688991e-03 [41] 4.502862e-04 1.252228e-05 7.881555e-02 3.142613e-02 4.846527e-03 [46] 2.562902e-03 4.846527e-03 1.101708e-03 7.252032e-02 2.205958e-02 [1] "Benjamini-Hochberg" [1] 1.000000e+00 1.000000e+00 8.940844e-01 8.510676e-01 1.000000e+00 [6] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [11] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [16] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 5.114323e-01 [21] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [26] 1.000000e+00 1.000000e+00 1.754486e-01 1.000000e+00 6.644618e-01 [31] 7.575031e-02 1.153102e-02 1.581959e-01 2.812089e-01 1.636176e-02 [36] 1.153102e-02 1.325863e-01 2.774239e-02 1.754486e-01 1.209832e-02 [41] 2.025930e-03 5.634031e-05 3.546073e-01 1.413926e-01 2.180552e-02 [46] 1.153102e-02 2.180552e-02 4.956812e-03 3.262838e-01 9.925057e-02 [1] "Benjamini & Yekutieli" [1] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [6] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [11] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [16] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [21] 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [26] 1.000000e+00 1.000000e+00 7.019185e-01 1.000000e+00 1.000000e+00 [31] 2.020365e-01 1.516674e-02 5.625735e-01 1.000000e+00 2.909271e-02 [36] 1.537741e-02 4.125636e-01 6.782670e-02 6.803480e-01 1.882294e-02 [41] 9.005725e-04 1.252228e-05 1.000000e+00 4.713920e-01 4.395577e-02 [46] 1.088915e-02 4.846527e-02 3.305125e-03 1.000000e+00 2.867745e-01 [1] "Bonferroni" [1] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [6] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [11] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [16] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [21] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [26] 9.991834e-01 9.991834e-01 4.632662e-01 9.991834e-01 9.991834e-01 [31] 1.575885e-01 1.383967e-02 3.938014e-01 7.600230e-01 2.501973e-02 [36] 1.383967e-02 3.052971e-01 5.426136e-02 4.626366e-01 1.656419e-02 [41] 8.825610e-04 1.252228e-05 9.930759e-01 3.394022e-01 3.692284e-02 [46] 1.023581e-02 3.974152e-02 3.172920e-03 8.992520e-01 2.179486e-01 [1] "Hochberg" [1] 9.991834e-01 9.991834e-01 9.991834e-01 9.987624e-01 9.991834e-01 [6] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [11] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [16] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.595180e-01 [21] 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [26] 9.991834e-01 9.991834e-01 4.351895e-01 9.991834e-01 9.766522e-01 [31] 1.414256e-01 1.304340e-02 3.530937e-01 6.887709e-01 2.385602e-02 [36] 1.322457e-02 2.722920e-01 5.426136e-02 4.218158e-01 1.581127e-02 [41] 8.825610e-04 1.252228e-05 8.743649e-01 3.016908e-01 3.516461e-02 [46] 9.582456e-03 3.877222e-02 3.172920e-03 8.122276e-01 1.950067e-01 Hommel
zkl
This work is based on R source code covered by the GPL license. It is thus a modified version, also covered by the GPL. See the FAQ about GNU licenses.
<lang zkl>fcn bh(pvalues){ // Benjamini-Hochberg
psz,pszf := pvalues.len(), psz.toFloat(); n_i := psz.pump(List,'wrap(n){ pszf/(psz - n) }); # N/(N-0),N/(N-1),.. o,ro := order(pvalues,True),order(o,False); # sort pvalues, sort indices in := psz.pump(List,'wrap(n){ n_i[n]*pvalues[o[n]] }); pmin := cummin(in).apply((1.0).min); # (min(1,c[0]),min(1,c[1]),...) ro.apply(pmin.get); # (pmin[ro[0]],pmin[ro[1]],...)
}
fcn by(pvalues){ // Benjamini & Yekutieli
psz,pszf := pvalues.len(), psz.toFloat(); o,ro := order(pvalues,True),order(o,False); # sort pvalues, sort indices n_i := psz.pump(List,'wrap(n){ pszf/(psz - n) }); # N/(N-0),N/(N-1),.. q := [1..psz].reduce(fcn(q,n){ q+=1.0/n },0.0); in := psz.pump(List,'wrap(n){ q * n_i[n] * pvalues[o[n]] }); cummin(in).apply((1.0).min) : ro.apply(_.get);
}
fcn hochberg(pvalues){
psz,pszf := pvalues.len(), psz.toFloat(); o,ro := order(pvalues,True),order(o,False); # sort pvalues, sort indices n_i := psz.pump(List,'wrap(n){ pszf/(psz - n) }); # N/(N-0),N/(N-1),.. in := psz.pump(List,'wrap(n){ pvalues[o[n]]*(n + 1) }); cummin(in).apply((1.0).min) : ro.apply(_.get);
}
fcn cummin(pvalues){ // R's cumulative minima --> list of mins
out,m := List.createLong(pvalues.len()), pvalues[0]; foreach pv in (pvalues){ out.append(m=m.min(pv)) } out
} fcn order(list,downUp){ // True==increasing, --> List(int) sorted indices
f:=(downUp) and fcn(a,b){ a[1]>b[1] } or fcn(a,b){ a[1]<b[1] }; [0..].zip(list).pump(List()).sort(f).pump(List,T("get",0))
}
fcn bonferroni(pvalues){ // -->List
sz,r := pvalues.len(),List(); foreach pv in (pvalues){ b:=pv*sz; if(b>=1.0) r.append(1.0); else if(0.0<=b<1.0) r.append(b); else throw(Exception.ValueError(
"%g is outside of the interval I planned.".fmt(b)));
} r
}
fcn hommel(pvalues){
psz,indices := pvalues.len(), [1..psz].walk(); // 1,2,3,4... o,ro := order(pvalues,False),order(o,False); # sort pvalues, sort indices p := o.apply('wrap(n){ pvalues[n] }).copy(); // pvalues[*o] npi := [1..].zip(p).apply('wrap([(n,p)]){ p*psz/n }); min := (0.0).min(npi); // min value in npi pa,q := List.createLong(psz,min), pa.copy(); #(min,min,,,) foreach j in ([psz - 1..2,-1]){ ij:=[0..psz - j].walk(); i2:=(j - 1).pump(List,'+(psz - j + 1)); q1:=(0.0).min((j-1).pump(List,'wrap(n){ p[i2[n]]*j/(2 + n) })); foreach i in (psz - j + 1){ q[ij[i]] = q1.min(p[ij[i]]*j) } foreach i in (j - 1){ q[i2[i]] = q[psz - j] } foreach i in (psz){ pa[i] = pa[i].max(q[i]) } } psz.pump(List,'wrap(n){ pa[ro[n]] }); // Hommel q-values
}</lang> <lang zkl>pvalues:=T(
4.533744e-01, 7.296024e-01, 9.936026e-02, 9.079658e-02, 1.801962e-01, 8.752257e-01, 2.922222e-01, 9.115421e-01, 4.355806e-01, 5.324867e-01, 4.926798e-01, 5.802978e-01, 3.485442e-01, 7.883130e-01, 2.729308e-01, 8.502518e-01, 4.268138e-01, 6.442008e-01, 3.030266e-01, 5.001555e-02, 3.194810e-01, 7.892933e-01, 9.991834e-01, 1.745691e-01, 9.037516e-01, 1.198578e-01, 3.966083e-01, 1.403837e-02, 7.328671e-01, 6.793476e-02, 4.040730e-03, 3.033349e-04, 1.125147e-02, 2.375072e-02, 5.818542e-04, 3.075482e-04, 8.251272e-03, 1.356534e-03, 1.360696e-02, 3.764588e-04, 1.801145e-05, 2.504456e-07, 3.310253e-02, 9.427839e-03, 8.791153e-04, 2.177831e-04, 9.693054e-04, 6.610250e-05, 2.900813e-02, 5.735490e-03);
bh(pvalues) : format(_,"\nBenjamini-Hochberg"); by(pvalues) : format(_,"\nBenjamini & Yekutieli"); bonferroni(pvalues) : format(_,"\nBonferroni"); hochberg(pvalues) : format(_,"\nHochberg"); hommel(pvalues) : format(_,"\nHommel");
fcn format(list,title){
print(title,":"); foreach n in ([1..list.len(),5]){ print("\n[%2d]:".fmt(n)); foreach x in (list[n-1,5]){ print(" %.6e".fmt(x)) } } println();
}</lang>
- Output:
Benjamini-Hochberg: [ 1]: 6.126681e-01 8.521710e-01 1.987205e-01 1.891595e-01 3.217789e-01 [ 6]: 9.301450e-01 4.870370e-01 9.301450e-01 6.049731e-01 6.826753e-01 [11]: 6.482629e-01 7.253722e-01 5.280973e-01 8.769926e-01 4.705703e-01 [16]: 9.241867e-01 6.049731e-01 7.856107e-01 4.887526e-01 1.136717e-01 [21]: 4.991891e-01 8.769926e-01 9.991834e-01 3.217789e-01 9.301450e-01 [26]: 2.304958e-01 5.832475e-01 3.899547e-02 8.521710e-01 1.476843e-01 [31]: 1.683638e-02 2.562902e-03 3.516084e-02 6.250189e-02 3.636589e-03 [36]: 2.562902e-03 2.946883e-02 6.166064e-03 3.899547e-02 2.688991e-03 [41]: 4.502862e-04 1.252228e-05 7.881555e-02 3.142613e-02 4.846527e-03 [46]: 2.562902e-03 4.846527e-03 1.101708e-03 7.252032e-02 2.205958e-02 Benjamini & Yekutieli: [ 1]: 1.000000e+00 1.000000e+00 8.940844e-01 8.510676e-01 1.000000e+00 [ 6]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [11]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [16]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 5.114323e-01 [21]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [26]: 1.000000e+00 1.000000e+00 1.754486e-01 1.000000e+00 6.644618e-01 [31]: 7.575031e-02 1.153102e-02 1.581959e-01 2.812089e-01 1.636176e-02 [36]: 1.153102e-02 1.325863e-01 2.774239e-02 1.754486e-01 1.209832e-02 [41]: 2.025930e-03 5.634031e-05 3.546073e-01 1.413926e-01 2.180552e-02 [46]: 1.153102e-02 2.180552e-02 4.956812e-03 3.262838e-01 9.925057e-02 Bonferroni: [ 1]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [ 6]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [11]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [16]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [21]: 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 1.000000e+00 [26]: 1.000000e+00 1.000000e+00 7.019185e-01 1.000000e+00 1.000000e+00 [31]: 2.020365e-01 1.516674e-02 5.625735e-01 1.000000e+00 2.909271e-02 [36]: 1.537741e-02 4.125636e-01 6.782670e-02 6.803480e-01 1.882294e-02 [41]: 9.005725e-04 1.252228e-05 1.000000e+00 4.713920e-01 4.395577e-02 [46]: 1.088915e-02 4.846527e-02 3.305125e-03 1.000000e+00 2.867745e-01 Hochberg: [ 1]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [ 6]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [11]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [16]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [21]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [26]: 9.991834e-01 9.991834e-01 4.632662e-01 9.991834e-01 9.991834e-01 [31]: 1.575885e-01 1.383967e-02 3.938014e-01 7.600230e-01 2.501973e-02 [36]: 1.383967e-02 3.052971e-01 5.426136e-02 4.626366e-01 1.656419e-02 [41]: 8.825610e-04 1.252228e-05 9.930759e-01 3.394022e-01 3.692284e-02 [46]: 1.023581e-02 3.974152e-02 3.172920e-03 8.992520e-01 2.179486e-01 Hommel: [ 1]: 9.991834e-01 9.991834e-01 9.991834e-01 9.987624e-01 9.991834e-01 [ 6]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [11]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [16]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.595180e-01 [21]: 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 9.991834e-01 [26]: 9.991834e-01 9.991834e-01 4.351895e-01 9.991834e-01 9.766522e-01 [31]: 1.414256e-01 1.304340e-02 3.530937e-01 6.887709e-01 2.385602e-02 [36]: 1.322457e-02 2.722920e-01 5.426136e-02 4.218158e-01 1.581127e-02 [41]: 8.825610e-04 1.252228e-05 8.743649e-01 3.016908e-01 3.516461e-02 [46]: 9.582456e-03 3.877222e-02 3.172920e-03 8.122276e-01 1.950067e-01