Error running EdgeR after StringTie

[simeix]

I’m having trouble running EdgeR after running StringTie. The error I get when I try to run is that the gene counts are differing in numbers. Prior to the update for Galaxy, we had no issues in running these samples, but now we’re unsure where the issues coming from.

How do we make it that when running the samples the gene counts numbers are the same? I ran StringTie with “User reference transcripts only” parameter as “Yes”, but the issue still stays.

[jennaj]

Welcome, @simeix

In general, we recommend using Featurecounts or HTseq-count instead for the counting step. An example of this usage can be found in a few of our tutorials.

• Discovery: Hands-on: De novo transcriptome reconstruction with RNA-Seq / De novo transcriptome reconstruction with RNA-Seq / Transcriptomics
• Known (fly): Hands-on: Reference-based RNA-Seq data analysis / Reference-based RNA-Seq data analysis / Transcriptomics
• Another known (human): Hands-on: 1: RNA-Seq reads to counts / 1: RNA-Seq reads to counts / Transcriptomics (plus the other two in the series)

That said, we have other reports of the newer Stringtie behaving differently, and that might be a problem, but I don’t have a concrete example yet. If you want to share your work I’m willing to take a closer look. Please see → How to get faster help with your question. I would need the shared history with everything in context for bug hunting.

Another solution could be to just keep using the version of the tool that already works for you.

So … whether the change is a Galaxy change (and maybe bug we can fix) or a change in the underlying tool itself (we can’t change that, only the Bioconductor team can do that) is not clear. And, we’d be interested in correcting anything that is possible on our side.

Let’s start there, thanks and hopefully we can follow up and solve this!

[simeix]

Thank you @jennaj !

In regards to HTseq-count, that has helped us with gene counts. We were wondering were there any alternative tools to use for calculating transcript counts other than StringTie? A similar error occurs with running EdgeR with transcript counts. Thank you for your help.

[jennaj]

Hi @simeix

EdgeR needs counts by gene, but DESeq2 can handle counts by transcript via TMP values (Salmon, Kalisto).

See the bottom of the DESeq2 tool for for examples, expected inputs, and links to tutorials with examples.

You can also search this forum with any of those tool names to learn more, and you can start up a new topic if you have some problem not discussed yet.

Thanks!