How does jgi_summarize_bam_conotig_depth handle paired reads?

[宋泽伟(Zewei Song)]

If we have paired reads in our alignment, will they be treated as two individual reads, or will they be grouped and considiered together?

I’m asking this because for some paired reads may have overlaps, while some have gaps. I am curious if these overlaps and gaps were considerred in your calculation.

Thanks!

Zewei

[Rob Egan]

each read is counted individually in the coverage for the region of the assembly that it aligns. if paired reads overlap the same region of the assembly, then their coverage would be counted twice for those bases. This is not ideal, but it is impractical to count the insert-sequence coverage instead of the read coverage for large assemblies in a limited amount of memory and a sorted BAM file.

Because overlapping read pairs may distort the coverage, it is almost always better to merge any paired reads that have an overlap and to use these merged reads for downstream coverage analysis. In general, you can get better assemblies using these merged reads also since there is precise insert-length information and error correction in the overlapping region.