Coverm filter inverse paired reads about coverm HOT 4 CLOSED

Hi,

Good point. The way coverm is interpreting this is that you want a BAM file with only proper pairs that don't pass the --min-read-percent-identity 0.95 --min-read-aligned-percent 0.7 thresholds. It doesn't "invert" the --proper-pairs flag.

I suggest taking a two step approach - grab reads that fail in the same way you did above, and then concatenate unpaired reads and unmapped reads with samtools and its flags for filtering.

Does that make sense? I understand this is a bit confusing, and the documentation could be better - will fix.
ben

from coverm.

Thanks for this quick and helpful response! I see, I think that makes sense. One follow up question, I guess I would then expect the bam file to still contain a matching number of reads in the forward and reverse file with this proper pairs flag, rather than having a different number between the files. I.e., if only proper pairs are passed to this output bam file, and the bam file contains output of properly paired reads which do not meet the identity and alignment requirements, should the fwd and rev fastq files have the same number of reads? Could you explain this piece a bit more?

from coverm.

Good q.

I think what is happening is that they are aligned as a proper pair, but then since you used thresholds which apply to a single side of each pair, then only one of the pair is present in the output BAM. Does that make sense?

I have to admit it has been a while since I looked at that code though. Perhaps it would be worth verifying by inspecting the mappings of now-singleton mappings in the original file?

from coverm.

I see, yes that does make sense. I'll definitely inspect the singletons to see if that is the case. Thanks!

from coverm.