Can somebody help me interper the results of my MultiQC? It seems like my lanes are...reverse and unstranded which sounds weird. Library used was Illumina Stranded mRNA Prep. From what I understood all reads from both lanes R1 and R2 in my case are coming from the same strand (apparently reversed) and they're both reverse? In the end even if the that metric if failed I have results that makes sense, right?

From what I understood all reads from both lanes R1 and R2 in my case are coming from the same strand (apparently reversed) and they're both reverse?

I think this is a misinterpretation. The R1 is always on the opposite strand as the R2. Depending on the library kit used (ask your sequencing lab about that), there are three possibilities: • The R1 corresponds to a random strand (coding or noncoding), because the kit is unstranded. This means that ~50% (give or take few because of sampling error) of reads are expected to match the coding strand. • The R1 always corresponds to the coding strand. • The R1 always corresponds to the noncoding stand. In the last two cases the kit is stranded, and ideally that means that typically ~98% of reads conform to the expectation of the strandedness -- if it is less than ~95%, either the kit if crap (but even in that case I would expect >90%) or something went wrong somewhere. In your case, it seems that while most reads are antisense/noncoding, a significant percentage (~20-30%) is not -- too many for a stranded kit, but not enough for an unstranded kit (that would be near 50%). This can happen in the case of DNA contamination with stranded kits (since DNA is inherently unstranded, and mixed with stranded mRNA the percentages are outside the expectations), but that does not seem to be the case here, because your exonic fractions are around 85%, which is about as good as you can expect (and they would be a lot lower with DNA). So I'm not really sure what is going on. I would take a look at the reads in IGV (you can color them by strand) and look through a few genes. What may be happening is that the weird distribution is limited to a few loci. Imagine you have a genes on one strand and a micro-RNA (which may regulate said gene) on the other, with both being expressed. If the miRNA is not in the GTF, it looks like all reads are from the gene itself, but with unexpected strandedness. If your reads of unexpected strandedness can be mostly explained by such cases, all is well -- just make sure the actual counting was done with "reverse" strandedness (you can look at the parameters for Salmon in the

execution_report_*.html

in the

script

column of the process table [make sure you look at the actual counting process, not the one run to determine strandedness; click on the column to expand]; worst case you have to rerun the pipeline with

reverse

strandedness for all samples). If, on the other hand, those reads are interspersed more or less evenly in all genes, my best guess is some weird issue with the kit.