Differential Gene Expression analysis using non-R package/tool

Question

[David Managadze]

I am trying to analyze Differential Gene Expression (DGE) in massive Counts files coming from RNA-seq and HTSeq-count. Doing this using DESeq2 takes very long time**. Also, R is not my language of choice.

What else is there to do DGE? I'd prefer the following (sorted by descending preference):

1. a native Python package (hopefully allowing parallelism)
2. executable/binary tool (maybe something more modern than old Tuxedo tools)
3. something else that does not internally call R packages e.g. DESeq2/EdgeR. One could call DESeq2 from Python using rpy2 but it would be slow.
4. Something that can employ power of Apache Spark cluster (Python/Koalas ?)
5. OK, be it DESeq2, but maybe parallelize it somehow :-)

Thanks for your help in advance!

Answer 1

[ATpoint]

People often use limma-voom when replicate numbers are large. I cannot find the thread now but I read that even the DESeq2 developer Mike Love recommends limma in that case.

Edit: Here is the thead I had in mind => https://support.bioconductor.org/p/73399/.

I think you are better of using R here because things are so established when using the typical Bioconductor packages. No need to change language. The manuals contain lots of example code you essentially only have to copy.

By the way, the most recent DESeq2 version had some code upgrades to better scale speed with sample size but as said limma is a good alterative.