LDmat

LDmat compresses and enables queryable storage of large linkage disequilibrium (LD) matrices in HDF5 format for population genetics analyses such as fine-mapping, LD score regression, and linear mixed models in GWAS.

Key Features:

• Compression (HDF5): Compresses extensive LD matrices into HDF5 file format to reduce storage while preserving matrix content.
• Query capability: Allows querying of compressed matrices to extract specific submatrices corresponding to genomic regions or selected loci.
• Flexible data extraction: Supports extraction of submatrices based on a provided list of loci or on loci filtered by minor allele frequency ranges.
• Reconstruction: Enables reconstruction of original file formats from compressed files to restore standard LD matrix representations.

Scientific Applications:

• Fine-mapping: Supports retrieval of LD submatrices for fine-mapping of genetic variants associated with traits or diseases.
• LD score regression: Facilitates LD-based inputs for LD score regression analyses relevant to polygenic architecture and polygenic risk estimation.
• Linear mixed models in GWAS: Provides LD data for linear mixed models used to account for population structure and relatedness in GWAS.

Methodology:

Implemented in Python; compresses LD matrices into HDF5 while preserving queryability, supports extraction of submatrices by specified loci or by minor allele frequency range, and enables reconstruction of original file formats from compressed files.