Is it better to pool samples prior to running deseq2, or should the algorithm be run separately for every pool?

Question

[Shelby]

When comparing results from running deseq2 separately, we saw that deseq does some pool-specific normalization. Is this desired or should we just group all data and then run it through deseq? All pools represent the same experiment. Seems like pooling all should be statistically more robust. Thank you!

If it makes a difference, we are using normalizationFactors instead of having the algorithm estimate sequencing depth.

Answer 1

[swbarnes2]

Seems like pooling all should be statistically more robust.

Yup. So put everything into the same DESeq2 object.

See:

https://bioconductor.org/packages/devel/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#if-i-have-multiple-groups-should-i-run-all-together-or-split-into-pairs-of-groups

[Shelby]

Gotcha, thank you!