ParascopyVC

ParascopyVC performs variant calling across low-copy repeats (LCRs) by aggregating reads mapped to all repeat copies and applying polyploid variant calling and paralogous sequence variant analysis to detect variants and estimate genotypes per copy.

Key Features:

• Multilocus Approach: Performs variant calling across all copies of LCRs simultaneously to accurately identify variants within repetitive regions, which are associated with genetic diseases when mutations occur in more than 150 overlapping genes.
• Utilization of Reads Independent of Mapping Quality: Aggregates reads mapped to different repeat copies and conducts polyploid variant calling without relying on mapping quality, enabling inclusion of ambiguously aligned reads.
• Identification of Paralogous Sequence Variants (PSVs): Identifies PSVs using population data to differentiate between repeat copies and support genotype estimation for each copy within LCRs.
• Superior Performance Metrics: On simulated whole-genome sequence data across 167 LCR regions it achieved precision 0.997 and recall 0.807, outperforming DeepVariant, GATK, and FreeBayes, and on the HG002 Genome-in-a-Bottle benchmark it achieved precision 0.991 and recall 0.909 across LCRs.
• Consistent Accuracy Across Human Genomes: Demonstrated higher accuracy with a mean F1 score of 0.947 across seven human genomes compared to the best competing method (F1 = 0.908).

Scientific Applications:

• Genetic research: Enables accurate identification of variants within LCRs to support studies of genetic predisposition and the molecular mechanisms of disease.
• Clinical diagnostics: Improves detection of pathogenic variants in repetitive genomic regions relevant to clinical genetic testing and variant interpretation.

Methodology:

Aggregates reads mapped to repeat copies, performs polyploid variant calling without relying on mapping quality, identifies PSVs using population data, and estimates genotypes for each repeat copy.