Quality control for GWAS studies

An important step in the analysis of genome-wide association studies (GWAS) is to identify problematic subjects and markers. Quality control (QC) in GWAS removes markers and individuals, and greatly increases the accuracy of findings.

Checking for gender (individuals whose genetic sex is discordant to their phenotypic gender), genotyping rate (call rate), minor allele frequency (MAF), Hardy-Weinberg equilibrium deviation (HWE), heterozygosity rate and identical by descent (IBD) allele sharing are useful QC steps. Standard tools like plink and king are called by the scripts.

The first step is to load VCF or BCF file into PLINK.

PLINK --bcf file.bcf.gz \
	--allow-no-sex \
	--keep-allele-order \
	--vcf-idspace-to _ \
	--make-bed \
	--out plink.load

Then remove SNPs with more than 10 percent missing genotype calls.

PLINK --bfile plink.load \
         --allow-no-sex \
         --keep-allele-order \
         --geno \
         --make-bed \
         --out plink.geno10pc

The command check-sex compares the sex reported in the .fam file and the sex imputed from the X chromosome inbreeding coefficients.

PLINK --bfile plink.geno10pc \
	--allow-no-sex \
	--keep-allele-order \
	--split-x hg38 \
	--make-bed \
	--out plink.geno10pc.split

PLINK --bfile plink.geno10pc.split \
	--allow-no-sex \
	--check-sex \
	--out plink.geno10pc.sexcheck

Complete pipeline can be found here. Now, you need to go into your SGE computer (our’s is called HGCC), and run:

qsub job_post_variant_qc.sh