BigLD

BigLD identifies haplotype blocks in dense genome sequencing datasets by clustering single nucleotide polymorphisms (SNPs) with strong pairwise linkage disequilibrium using interval graph modeling to improve representation of genetic variation and alignment with recombination hotspots.

Key Features:

• Interval Graph Modeling: Employs interval graph modeling to identify and cluster SNPs into LD bins based on pairwise linkage disequilibrium, including SNPs that are not physically contiguous.
• Agglomerative Approach: Begins with small SNP communities within each LD bin and progressively merges them to form larger haplotype blocks.
• Maximum-Weight Independent Set Method: Uses a maximum-weight independent set algorithm to determine the optimal number of LD blocks.
• Improved Block Size and Recombination Hotspot Alignment: Produces larger LD blocks that show higher agreement with recombination hotspot locations identified by sperm-typing experiments compared to MATILDE, Haploview, MIG++, and S-MIG++.
• Efficiency and Performance: Processes 13,288,240 non-monomorphic SNPs from the 1000 Genomes Project autosome data for 286 East Asians in approximately 5.83 hours, demonstrating computational efficiency relative to previous methods.

Scientific Applications:

• Multi-SNP-based Association Analyses: Specifies SNP sets for multi-SNP association testing.
• Haplotype Block Identification: Identifies haplotype blocks within high-density sequencing data.
• Recombination and Linkage Analysis: Aligns LD block boundaries with recombination hotspots to inform linkage and recombination studies.
• Population Genetics and Genetic Epidemiology: Supports analyses of population structure and genetic epidemiology.
• Hereditary Disease and Evolutionary Biology Studies: Facilitates investigation of genetic linkage patterns relevant to hereditary diseases and evolutionary processes.

Methodology:

Uses interval graph modeling to form LD bins, an agglomerative clustering strategy that merges small SNP communities into larger haplotype blocks, and a maximum-weight independent set method to select the optimal set of LD blocks.