Low assignment rate with featureCounts

Question

[gt]

I used STAR to align my reads (brain samples) to human reference genome. Getting good unique mapping rates (~70-90%).

However, when I use featureCounts I get really low assignment rates.

Here is an example command

featureCounts -p -t exon -g gene_id -s 2 -T 16 -a gencode.v38.annotation.gtf -o counts.txt Aligned.sortedByCoord.out.properPairs.bam

Any ideas why my assignment rate is so low? Should I be mapping to genes instead? Even if I do that I only get around 26% assignment which is better but still very low.

[Mensur Dlakic]

Have you tried leaving out the -s 2 switch? That and -t gene should give you higher mapping fraction.

[gt]

Yes, I have done both os those things. Leaving out -s actually gives me a lower mapping rate because the reads a stranded. I have also tried -t gene and it only results in slightly higher mapping rates.