1
$\begingroup$

I need to identify differential genes across 8 groups. I know this can be done using DESeq2 or EdgeR as these are better suited and equipped.

However, I was thinking if something like this piece of code could be tweaked. Here it works for two groups each with 3 replicates and generates a heatmap. The code is in R

pvals <- apply(dm, 1, function(x) wilcox.test(x~c(rep(0,3), rep(1,9)))$p.value)
heatmap(dm[pvals<0.05,])

Thank you for your inputs and suggestions.

$\endgroup$
  • 5
    $\begingroup$ What is your reason not to use DESeq2 or EdgeR? Their models will work much better for RNAseq data than a simple Wilcoxon test. $\endgroup$ – PPK Aug 14 '19 at 15:11
  • $\begingroup$ Welcome to Bioinformatics.SE! At the moment, your question simply asks "what about this?" while ignoring a tremendous amount of scholarship on differential expression analysis. Unless you can more clearly present a specific question or concern, there is a good chance this question will be closed as too broad or unclear. $\endgroup$ – Daniel Standage Aug 14 '19 at 20:06
  • $\begingroup$ For a single measurement of some fixed, observable quantity like height or grades, using a Wilcoxon test isn't a bad idea. But RNA-seq is a very complicated system and can't simply be analyzed by something like a Wilcoxon test. Your results will be full of mistakes if you do this. I would recommend that you read some introduction to RNA-seq papers or textbooks and watch some presentations by Wolfgang Huber, Gordon Smyth, and Lior Pachter to see why this is the case $\endgroup$ – James Hawley Aug 16 '19 at 15:26