I was wondering if somebody could explain to me how featureCounts differentiates (or not) between reads coming from the foreward or reverse strands in regards to overlapping antisense genes.
From my understanding, it tosses out reads that overlap two features. Therefore, I tried to use the strandSpecific=1 option, even though my rna-seq protocol is not strand-specific. Predictably, the results make little sense. However, I can split the .bam files manually into + and - strands using samflags (and the resulting tracks look good).
How does featureCounts differentiate between the different strands? Do I have an easier option to get an accurate count table than splitting the bam files manually and quantifying the counts for both strands seperately? My annotation file is the mm10 GENCODE gtf.