Would I want to call variants on hs37d5 for any reason?

Question

[Manuel]

Of course, one should use hs37d5.fa for read mapping.

However, should I bother to call variants on the decoy contig hs37d5? GATK Haplotype Caller is spending ages on the contig and I will never use the variant calls.

Would this contig be of any use downstream? Would it make sense to use the GATK variant recalibration on the variant calls of this contig or could they be safely ignored?

Thanks!

Answer 1

[mjaeger]

Hi,

we included hs37d5 for the mapping step to get rid of mismapped reads and to increase the SNV-calling sensitivity. See also GATK forum decoy discussion

For downstream analysis we ignore scaffolds and contigs, since most databases and tools are unaware of these.