Closed erzakiev closed 1 year ago
Hi', same doubt but about results obtained with edgeR: i have logFC, logCPM, F, Pvalue and FDR as columns of DEGs analysis.
Which one could be used instead of t
?
Hi @erzakiev and @ultimatex5,
You can use the differential expression analysis framework that you prefer. decoupleR in the end requires a gene level statistic to perform enrichment analysis but it is agnostic of how it was generated. However, we do recommend to use statistics that include the direction of change and its significance, for example the t-value obtained for limma or Deseq2 (stat
). The problem with edgeR is that it does not return such statistic but you could create your own by weighting the obtained logFC by pvalue with this formula: -log10(pvalue) * logFC
.
I will update the vignettes to clarify this point, thanks for pointing this out. Hope this is helpful! Let me know if you have more questions.
Thank you, looks ok for me, @PauBadiaM !
@ultimatex5 I'll close the issue if you don't have any remaining Q regarding edgeR
?
You can close. Thank you!
In the transcription factor activity vignette the method of choice for calculating differentially expressed genes is limma.
And in the end we only use the
t
statistic for the TF inference.stat
for that purpose instead, like the one found in theresults
output of its analysis?Example of a DESeq2 result: