cannot compute log geometric means Deseq2

[AnjaliC4]

Hi,
Thank you for this great package. I am using Deseq2 mthod with pbDS function and got this error:

Error in estimateSizeFactorsForMatrix(counts(object), locfunc = locfunc, :
every gene contains at least one zero, cannot compute log geometric means
Calls: pbDS ... tryCatch -> tryCatchList -> tryCatchOne ->
Execution halted

It seems like this error pops up in Deseq2 if every sample has atleast 1 zero. I do not want to add a pseudocount to all my values. Is there a way around this in Muscat ?

Thanks,
Anjali

[HelenaLC]

Just out of curiosity (to see whether this is filtering-related), could you try 1) running with method = "edgeR" and 2) funning with min_cells = 0, filter = "none" (or different variations of filter)?

[AnjaliC4]

Hi @HelenaLC you are correct. It is indeed related to the filtering issue. When I say min_cells=10 with DESeq2 - it works and min_cells=5 with edgeR also works. There is an option in Deseq2 to use -> dds<-estimateSizeFactors(dds, sfType = 'iterate' or 'poscounts') which is specifically made to handle this error case. Could you please enable choosing 'type' option pbDS function with method DESeq2? Is this something that can be quickly implemented, if possible?