Get genes and coverage/depth per gene from sam/bam

Question

[prishly]

Hi, Is there any tool (preferably one that could be run locally) that gets genes and coverage/depth per gene from sam/bam file (like sum of reads that map to given gene)? I mean coverage per gene, not position/locus, with no prior knowledge of what genes to look for (no bed file).

[WouterDeCoster]

And how is that tool supposed to delineate genes, then?

[h.mon]

There isn't, but this is a very good idea for a machine learning method.

[Longshotx]

Any luck with this @prishly?

Answer 1

[Charles Warden]

I'm not sure what issues there could be with looking at coverage when the de novo annotations are from the same sample (compared to an independent set of annotations), but you could theoretically generate regions with derfinder (or do a reference-based de novo assembly, with a program like cufflinks).

Once you had your regions, you could use RSeQC to generate coverage plots.