BQSR using GATK without dbsnp

Question

[kezcleal]

Hi, I have run the base quality score recalibration (BQSR) tool from GATK but for the known sites parameter I used only two databases:

• Mills_and_1000g_gold_standard_indels
• 1000G_phase1.indels

I realise I should possibly also have included the dbsnp_138 database in this list. Is it necessary to re-run with all three databases?

[Ashutosh Pandey]

Ideally, it would be better if you can re-run the BQSR analysis using both SNPs and Indels. But it may or may not affect the downstream variant results depending on how good your sequencing data is. Honestly speaking, I have never found a big difference in the number and quality of variants called before or after BQSR but it could be solely due to genuinely good quality of reads that we get from our sequencing core. The other steps such as indel-realignment do have a big effect on the quality of variants.

Answer 1

[youcai]

I'm working on a genome that doesn't really have a good SNP database yet. I'm wondering if it still makes sense to run base quality score recalibration without known SNPs