You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In version 3.3.1 of Goby, I see gominer.Fisher is used to compute p-values of Fisher tests in org/campagnelab/goby/stats/FisherExactTestCalculator.java.
During the test org.campagnelab.goby.stats.TestStatistics.testTwoStats, the p-value of the Fisher test associated to the table
322777 318050
160397 151279
is computed, and the p-value given by gominer.Fisher is 1. Yet, passing this table to R (fisher.test) or even to https://www.langsrud.com/fisher.htm gives a p-value that is lower than 1.E-16. So the test testTwoStats should be changed as it should not expect a p-value of 1 for the Fisher test.
Enclosed is a patch which would avoid needing gominer and use DistLib (https://sourceforge.net/projects/statdistlib/) instead. It needs a review before maybe being included, still.
Hello,
In version 3.3.1 of Goby, I see gominer.Fisher is used to compute p-values of Fisher tests in org/campagnelab/goby/stats/FisherExactTestCalculator.java.
During the test org.campagnelab.goby.stats.TestStatistics.testTwoStats, the p-value of the Fisher test associated to the table
322777 318050
160397 151279
is computed, and the p-value given by gominer.Fisher is 1. Yet, passing this table to R (fisher.test) or even to https://www.langsrud.com/fisher.htm gives a p-value that is lower than 1.E-16. So the test testTwoStats should be changed as it should not expect a p-value of 1 for the Fisher test.
Enclosed is a patch which would avoid needing gominer and use DistLib (https://sourceforge.net/projects/statdistlib/) instead. It needs a review before maybe being included, still.
computing_fisher_pvalue_hypergeom.txt
Best regards,
--
Pierre
The text was updated successfully, but these errors were encountered: