Genotype calling and phasing using next-generation sequencing reads and a haplotype scaffold

Area of Science:

• Genomics
• Bioinformatics
• Statistical Genetics

Background:

• Large-scale genomic studies often use low-coverage sequencing due to cost.
• Leveraging linkage disequilibrium is key for genotype inference from low-coverage data.
• Existing methods can be enhanced by integrating genome-wide microarray genotypes.

Purpose of the Study:

• To propose and evaluate a novel method (MVNcall) for genotype calling and phasing.
• To utilize both low-coverage sequencing data and genome-wide microarray genotypes.
• To improve imputation accuracy, especially for rare variants.

Main Methods:

• MVNcall employs an MCMC algorithm to phase polymorphic sites.
• It uses genotype likelihoods and local haplotype information for allele updates.
• A multivariate normal model captures allele frequency and linkage disequilibrium.
• Mendelian transmission constraints are incorporated when trio data is available.

Main Results:

• The method was evaluated using Phase 1 of the 1000 Genomes Project (1KGP).
• Performance was assessed based on genotype accuracy, phasing accuracy, and imputation.
• Inferred African haplotype panels boosted rare variant imputation accuracy (R2 >0.05 for MAF <1%).

Conclusions:

• Integrating microarray genotypes with sequencing data enhances variant calling.
• MVNcall improves imputation accuracy, leading to increased power for association studies.
• The method is highly parallelizable and valuable for large genomic datasets.