Alternate-locus aware variant calling in whole genome sequencing.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Human genome assemblies have evolved from linear to graph-like models to better represent structural variability.
• Graph-based models offer improved accuracy for variant calling in whole-genome sequencing (WGS).

Purpose of the Study:

• To develop an algorithm for analyzing variant call patterns in structurally variable regions of the GRCh38 human genome assembly.
• To improve the technical validity of variant calling in WGS by distinguishing between primary and alternate loci.

Main Methods:

• Developed an algorithm to analyze variant call patterns in 178 structurally variable regions of GRCh38.
• Inferred sample sequence origin (primary, alternate, or heterozygous) for these regions.
• Applied the algorithm to 121 in-house WGS datasets aligned to GRCh37 and GRCh38.

Main Results:

• Identified "alignable scaffold-discrepant positions" (ASDPs) where sequence identity leads to spurious variant calls.
• Found that an average of 51.8% of 178 regions best corresponded to an alternate locus.
• Filtered genomes using the algorithm, identifying 7863 variant calls per genome at ASDPs and finding 437 genome-wide association study hits at ASDPs.

Conclusions:

• The algorithm effectively avoids spurious variant calls by utilizing information from GRCh38's structurally variable regions.
• Incorporating graph-like genome assembly resources holds significant potential for variant calling.
• Development of computational resources is crucial for full genotype reconstruction in personal genomes.