TimelyScholarFree Tools › FDR Calculator (Benjamini-Hochberg)
Correct for multiple testing. Paste a list of p-values to get Benjamini-Hochberg (FDR), Benjamini-Yekutieli, Holm or Bonferroni adjusted p-values, with the tests that remain significant.
FreeNo account neededYour data stays in your browser
| Significant after Benjamini-Hochberg (FDR) | 5 of 10Adjusted p ≤ 0.05 |
|---|---|
| Significant before adjustment | 7 of 10 |
| Rank | Test | p | BH critical value (i/m)q | Adjusted p | Significant |
|---|---|---|---|---|---|
| 1 | gene_A | 0.0004 | 0.0050 | 0.0040 | Yes |
| 2 | gene_B | 0.0030 | 0.0100 | 0.0150 | Yes |
| 3 | gene_C | 0.0110 | 0.0150 | 0.0367 | Yes |
| 4 | gene_D | 0.0190 | 0.0200 | 0.0475 | Yes |
| 5 | gene_E | 0.0240 | 0.0250 | 0.0480 | Yes |
| 6 | gene_F | 0.0410 | 0.0300 | 0.0683 | No |
| 7 | gene_G | 0.0490 | 0.0350 | 0.0700 | No |
| 8 | gene_H | 0.1200 | 0.0400 | 0.1500 | No |
| 9 | gene_I | 0.2800 | 0.0450 | 0.3111 | No |
| 10 | gene_J | 0.4600 | 0.0500 | 0.4600 | No |
FDR methods control the expected proportion of false discoveries among the results called significant; FWER methods (Holm, Bonferroni) control the chance of any false positive and are more conservative. Benjamini-Yekutieli is valid under any dependence between tests.
One per line, optionally with a label such as a gene name.
For example Benjamini-Hochberg at q = 0.05.
Sorted table with adjusted p-values.
When many hypotheses are tested, some will be significant by chance. Family-wise error methods such as Bonferroni and Holm control the chance of even one false positive and become very strict with many tests.
The Benjamini-Hochberg procedure instead controls the false discovery rate: the expected proportion of false positives among the results declared significant. It is widely used in genomics and other studies with many tests. The Benjamini-Yekutieli variant remains valid under any dependence between tests.
Use Bonferroni or Holm when any false positive is costly and tests are few; use Benjamini-Hochberg for exploratory analyses with many tests.