Metastats q < p ?

mothur bugs
1

Hi!

For the q-values in metastats are a correction for multiple testing I would expect q > p. But I have got some metastats files containing both q > p and q < p. Moreover, some files even show all the same negative q-value although the p-value seems fine.
I have used v.1.33.3 (and also tried v.1.34.4), there were no error messages, and the shared file looks quite normal.

Thank you for any suggestions!
Michael

2

The q-value is based on a curve fit to the p-values and can generate some screwy values. You might just take the p-values and apply your own correction for multiple comparisons.