goseq error in length and/or count files?

[mars142]

Hello there,
I did some RNAseq analysis, which actually worked well.
Now I wanted to move on with DEG analysis. I took the result files from the former analysis and everything worked fine until the goseq analysis:

Error in [.default(summary(map), , 1) : incorrect number of dimensions

This happens either I manipulate my data or use the originals from the former feature count analysis.

I know that two coworkers also used these exact workflows without any issues.
May you help with some expertise? Thanks

History:

Galaxy

Workflow invocation: 9dbd24ecdd3fcec8

[jennaj]

Welcome @mars142

Thanks for sharing your history, super helpful!

The problem is with the Ensembl gene identifiers – these have an extra version .N and the goseq tool doesn’t understand how to parse those correctly.

If you scroll down on the tool form, you’ll see the expected format, and those happen to use Ensembl too.

I’m guessing that your coworkers were either using a different annotation source, or they removed the versioning.

This recent topic here has more details (a different Bioconductor tool, but all are in R and work the same at a technical level). →

@jennaj

Some of the Bioconductor tools do not understand the “dot” in identifiers. Sort of a gotcha but that is how the tools work everywhere. In short, R is interpreting the dot when it shouldn’t be. I think you can quote it but I haven’t tested that with every tool in this pipeline.

We have a few topics about it if you are curious. The solution can be to remove the .N part of the identifier. This is one example.

• how to replace these ID with official gene names?

The FAQ here has a troubleshooting warning but it is easy to miss!

• FAQ: Extended Help for Differential Expression Analysis Tools
  • Sometimes these tools do not understand transcript_id.N and gene_id.N notation (where N is a version number).
  • This notation could be in fasta or tabular inputs.
  • Try removing .N from all inputs, and check for the accidential creation of new duplicates!

This trips up people using the tools directly, too!

• https://support.bioconductor.org/post/search/?query=version+numbers+on+identifiers

---

What to do

• Add a step into the workflow that strips of the .N content in the tabular files before sending the data into goseq.
• Find an annotation file that already has the version stripped.
• More about genome data sources. → Reference genomes at public Galaxy servers: GRCh38/hg38 example

Hope this helps!

[mars142]

Hey, thanks for your help. Seemed to work. Now I just got the problem of having absolute zero differences in expression values, which should not be the case. I can not find the case for that. Maybe I switched data or parameters. But I did not find any differences, while going through the training material.
For your information, I want to analyse 5 treated cell cultures against one control. So looking over all experiments, that were done before, there must be a difference in gene expression.

Galaxy

[jennaj]

Hi @mars142

It is hard to see what is going on with this result. But I do see that you had problems linking back in the annotation. You could switch to using the UCSC version of this annotation instead. It already has the simplified Ensembl genes/transcripts and will work with all these tools without extra manipulations.

• https://hgdownload.soe.ucsc.edu/goldenPath/hg38/bigZips/genes/

Other than that, maybe only having one control sample is the root issue.

There is probably a lot of discussion about this online, but we have one captured in our FAQ here as a starting place.

• FAQ: Extended Help for Differential Expression Analysis Tools
• See → Differential expression tools all require sample count replicates. Rationale from two of the DEseq tool authors.

Hope this helps!