Hi everyone,
I am comparing different proteomes from fractionation experiments. No SILAC was used, only LFQ. I started to analyze the data with the DEP package, only to find out that the adjusted p.value were very high for almost all proteins. The original paper, where the package was developed, says "the fraction of differentially enriched proteins should not exceed 10–15% of the quantified proteins". In my data, the different samples overlap, but the "differential expression" of my proteins is higher than 15%.
1) Is there a way to change the priori assumption that the majority of proteins are not differentially enriched?
2) Can I use the p.value returned by the F statistic of the DEP package and calculate the adjusted p.value using the Benjamini-Hochberg FDR test? Would that make sense?
Thanks a lot, Rob
I am the limma author but I can't answer questions about DEP.
There is no requirement in limma that DE be less than 15%.
If you have p-values from limma you can apply Benjamini Hochberg by using
p.adjust(p, method="BH").Hi Gordon, DEP is built ontop of limma. for some reason, the authors decided to compute adjusted p-value using fdrtool from the t-statistics returned by limma. https://github.com/arnesmits/DEP/blob/b425d8d0db67b15df4b8bcf87729ef0bf5800256/R/functions.R#L951 Maybe you could answer in the view of a limma computation only, and that would be helpful. Best.