I'm using OUTRIDER, which I understand is very similar in terms of initial data object creation to deseq2. the OUTRIDER datasets are the same in structure as deseq datasets.
In my analysis, I'm receiving the following error at some stage
"Error in estimateSizeFactorsForMatrix(fcMat) : every gene contains at least one zero, cannot compute log geometric means"
However, this is not the case, I've looked manually and there are many genes which don't have zero counts for a sample. Biologically it would be impossible for ALL genes to have at least one zero count as some are necessary for life. My best guess it somehow when reading in to creaste an OUTRIDER dataset (again, like the deseq dataset), a column is somehow misaligned or something, and appears to have zero values. But the notion that every gene could have at least one zero counts is impossible biologically. So I don't want to force functionality by adding a pseudocount of 1 to everything. Something systemic is wrong.
In the counts table text file all is fine. When I load into an OUTRIDER data set (much like deseq data set) all seems ok , then I run an analysis step , and the problem occurs. How can i view the deseq dataset, gene by gene to see where there are zero counts in highly expressed genes and subsequently perform better diagnostics on the problem.
It's hardly impossible if you have some failed samples with hardly any reads.
Please add a minimal reproducible example and code.