DESeq2 input is non-featurecounts

[jennaj]

Welcome, @CarinaS

For this part of your question

@CarinaS

Can one use DESeq2 by using count files from another source than featurecounts? I have obtained sequencing data from a company that included counts, in tabular format (first column is geneID, second column counts)

Yes, the counts can be generated from many different tools. Some in Galaxy as examples are Featurecounts and HTseq-count but the tool can also interpret TMP values from Salmon, Sailfish and Kallisto. You can also use Stringtie sometimes if the data is organized correctly.

You can also use tools like EdgeR and Limma to perform expression analysis. And many others…

@CarinaS

When I try this in DESeq2 I get an error about size factor, I have checked to have genes that have non-zero counts, so this shouldn’t cause a problem, I think.

That could be an actual scientific result, or a technical problem is leading to a spurious error.

These tools work in Galaxy the same as they do anywhere else, so sometimes searching with an error message at their forum can be informative. But you should probably always come back here to the Galaxy support site with new questions.

• Bioconductor - Support Site

@CarinaS

Can I continue with the counts I have, or should I try to repeat the alignment and then use the featurecounts files?

Without seeing your data and the error, this is difficult to answer. But in general, having more control over the data provenance will yield better results, both in how those are generated and how much you can “trust” them.

This guide in our FAQs explains about the technical bits to consider when troubleshooting.

• FAQ: Extended Help for Differential Expression Analysis Tools

For how these tools actually work from a scientific perspective, the Bioconductor site above and public scientific forums such as Biostars.com are good resources. The Galaxy tool form attempts to summarize some of this with link outs to publications, author websites and GTN tutorials that include that particular tool.

• GTN tutorials for DESeq2: Determines differentially expressed features from count tables

These are good questions so I’m being a bit verbose to create a mini-guide here in your topic for anyone else reading later on, too.

What to do

1. Check for the technical issues in our guide above. Can you notice and correct anything obvious?

   • If you get stuck, you can post back a share link to your history for feedback. How to is in the banner of this forum, also here → How to get faster help with your question

2. If you have the reads, maybe also consider generating the data that way through one of the tutorial workflows.

   • This would give you another set of results to compare to the counts you already have, to help you make decisions about how to interpret both sets of data.
   • Counts are based on reference data – and different sources and versions of references can differ in important ways.
   • Then how those counts are generated can differ in scientifically important ways: comparisons to only known artifacts versus known+discovery is a common way to approach this.

Let’s start there, thanks!