Why edgeR ouput some p values larger than 1

idyll_ty · 01-04-2012, 10:55 PM

I was using edgeR to do differential expression on a RNA-seq dataset, but when I check the p.value distribution of the output, I found ~700 genes got a p value larger than 1... Just wondering if anyone else got this problem before? And why I got such a weird a result. Here's my codes,

d <- DGEList(counts=countsTable, group=conds, lib.size = lib.sizes) #I used the raw read counts as instructed
d <- calcNormFactors(d, method="TMM")
d <- estimateCommonDisp(d)
et.commonDis <- exactTest(d)

hist(et.commonDis$table$p.value, breaks=100, col="skyblue", border="slateblue", main="pval distribution_edgeR")

I got a picture as attached...

May I have your ideas? Thanks.

Gordon Smyth · 02-01-2013, 04:14 PM

edgeR doesn't produce p-values greater than one. If you did observe this, then it would be a bug in the software, and you should write to the authors, who always appreciate being alerted to such things. But make sure first you are using the current software and that you can reproduce the problem.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
assembly with unsatisfying results - use new reads with larger inserts?	martin_313	Bioinformatics	4	01-23-2012 01:52 AM
Cuffdiff - ouput problem	ytmnd85	Bioinformatics	12	12-02-2011 09:10 AM
edgeR separate offset values/normalizations for each tag	sruddy	RNA Sequencing	1	07-08-2011 04:59 PM
samtools (or other tools) to find larger indels	waalkes	Bioinformatics	0	06-15-2011 03:53 PM
Bowtie ouput in Tophat	unidodo	Bioinformatics	2	03-30-2011 01:13 PM

02-01-2013, 04:14 PM	#2
Gordon Smyth Member Location: Melbourne, Australia Join Date: Apr 2011 Posts: 91	edgeR doesn't produce p-values greater than one. If you did observe this, then it would be a bug in the software, and you should write to the authors, who always appreciate being alerted to such things. But make sure first you are using the current software and that you can reproduce the problem.