Incorrect Group ID assigned by featurecounts_deseq2.r

[ikeller]

featurecounts_deseq2.r expects that sample names end in "_R1" etc and will trim the last 3 characters of the sample name to obtain the group ID for DESeq2. This will not work when there are 10 or more replicates per group. In this case, you end up with twice the expected number of experimental groups, because the underscore is retained in some replicates, for example: Tissue1 and Tissue1_. It would be safer to remove everything starting at the underscore.

[drpatelh]

Yes, this is a bug @ikeller 👍 I could swear I fixed this already but I obviously havent!

atacseq/bin/featurecounts_deseq2.r

Line 81 in 818c8dc

groups <- substr(samples.vec, 1, nchar(samples.vec)-3)

The easiest and safest solution would be to split on _ and then join everything back together except the last item.

Is this the only step that is failing to deal with the double-digit replicate number?

Lucky you for having more than 9 replicates! 😆

[ikeller]

Thanks for your quick reply.

I haven't spotted any other issues caused by the 2-digit replicate numbers.

[drpatelh]

No worries 👍 Ill try and fix this in the next release which will hopefully be in a couple of weeks.

[drpatelh]

Fixed here @ikeller so will be sorted in the next release.