Hello
I recently got a big RNAseq analysis to do downstream analysis and comparison with other
published experiments. The group analyzed the data with the online tool DAVID for gene ontologies and got very excited about some of the enriched GO Terms.
I wanted to repeat part of the analysis prior any further comparisons and found in the DAVID output files some strange results.
The problem is that even though a Benjamini pvalue of <= 0.05 was used to filter significant results the field FDR (false discovery rate) can be pretty high often exceeding 0.05 and sometimes grow as large as e.g. >1.5 .
Now please correct me if I am wrong, but the way I thought this would be structured is that
the normal p-value represents the potential significance within the GO Term and the adjusted Benjamini should take into account the potential FDR.
Does anybody know more about this ?
Cheers
I recently got a big RNAseq analysis to do downstream analysis and comparison with other
published experiments. The group analyzed the data with the online tool DAVID for gene ontologies and got very excited about some of the enriched GO Terms.
I wanted to repeat part of the analysis prior any further comparisons and found in the DAVID output files some strange results.
The problem is that even though a Benjamini pvalue of <= 0.05 was used to filter significant results the field FDR (false discovery rate) can be pretty high often exceeding 0.05 and sometimes grow as large as e.g. >1.5 .
Now please correct me if I am wrong, but the way I thought this would be structured is that
the normal p-value represents the potential significance within the GO Term and the adjusted Benjamini should take into account the potential FDR.
Does anybody know more about this ?
Cheers
Comment