Good evening, I am doing a DE analysis on rna-bulk seq data aligned with salmon (with DESeq2 standard workflow). Recently, I found that, biologically, a particular gene is important/relevant for the analysis (by experiments performed). I checked and it is not Differentially Expressed. Morehover, I found that it is actually present in the data extracted right after the salmon alignment, but all the reads are zero (for each samples). Thus, when the pre-filtering is applied, the row with the gene is correctly deleted from the set.
It can be possible that an error occurred in the salmon alignment ? if yes, how can I check for inconsistencies ? I thought that having all reads equal to zero, for each sample, in unlikely to be an error. But I don't know if this assumption is correct.
Thank you for the help.
Do I understand correctly that you have one gene you are interested in, but all counts are zero, right? If so, first thing could be to use an alternative aligner, for example STAR, followed by featureCounts, and see how counts come out. Or visualize the STAR bam file on the Integrated Genome Viewer. That would exclude a software-specific issue. If the gene is difficult to map then one software might to better/worse than another, given in-built heuristics and cutoffs. It can of course be that expression is quite low, and you don't pick up the gene at current sequencing depth, or that (for whatever reason) the gene is simply not captured (given it's expressed) in the assay. After all, absence of counts is not evidence for no expression. Can you add some details why you think it should be detected?
Thank you for the suggestions. Unfortunately I cannot add any details for lack of information. But I will try using a different alignment and compare it to salmon results.