hs37d5 samtools idxstats

Question

[IRECG]

Hello, I am using samtools idxstats and I'd like to know what is hs37d5. Is that only the decoy sequences or the sum of reference genome + decoy sequences?

I mean, I have different BAM files and I want to have a measure of the "coverage" of each one, I thought about calculating the percentage of mapped reads (= #mapped reads ÷ sequence length) using samtools idxstats output. Can I use that or is not representative and should I use the #mapped reads from chr1-chrY?

Thanks!

Answer 1

[Kevin Blighe]

Quite a few resources from reputable individuals (including our very own Devon) and companies on this:

• Why Is There A Sequence With Name ">Hs37D5" In The Reference Hs37D5.Fa File Of Gatk Resource
• Google Genomics
• Towards Better Understanding of Artifacts in Variant Calling from High-Coverage Samples (Heng Li)
• Which human reference genome to use? (Heng Li)

That should firmly help to explain hs37d5.

----------------------------------------

For 'coverage' metrics, you appear to be referring to depth of coverage, which would be better served by using BEDTools coverage or BEDTools genomecov. SAMtools idxstats will give a different type of information, i.e., number of reads aligning to each of your contigs, and is more used for gauging overall alignment metrics.

Kevin

[IRECG]

Thank you! I haven't done BEDtools genomecov or coverage because in both cases I need my BAM, but also another file : -g in the first case and -a in the second one. I guess that in both cases it must be my reference genome, am I right?

[Kevin Blighe]

Yes, take a look here, where even the developer of BEDTools has provide an answer: Bedtools genomecov to identify regions at any coverage