A question about differential expression analysis

Question

[tunl]

My understanding is that most differential expression analysis tools for RNA-seq (bulk or single-cell) use negative binomial distribution, while some use the likelihood ratio test.

Would it be possible to do a simple t-test for differential expression analysis of single-cell RNA-seq data?

Thank you very much for your advice!

Answer 1

[Devon Ryan]

For clarification, most tools assume counts follow a negative binomial distribution and use either a likelihood ratio test or a Wald test.

Regarding using a simple T-test, DO NOT DO THIS. You do not have the knowledge required to correctly implement custom statistical methods on datasets like this. If you prefer a T-test, you can use limma/voom, which uses a moderated T-statistic.

[theobroma22]

There are so many good tools for scRNA out there to use!

[Devon Ryan]

Yup, from traditional RNAseq packages (DESeq2, edgeR, limma, etc.) to scRNAseq-specific packages, there's certainly no dearth of choices. I have yet to see a single "best in practice" package, but then again looking at DE genes is generally not the purpose of scRNAseq.

[tunl]

Thank you so much Devon! This is great info and advice!

[WouterDeCoster]

If an answer was helpful you should upvote it, if the answer resolved your question you should mark it as accepted.