JAWAMix5

JAWAMix5 performs out-of-core association mapping of high-throughput sequence-derived genotype data using HDF5-backed storage to enable scalable genetic association analyses.

Key Features:

• Out-of-core HDF5 storage: Stores genotype data on disk with HDF5 and accesses it as if in main memory to enable analyses beyond available RAM.
• Analytical functions: Implements standard linear models, linear mixed models, rare variant testing, analysis for nested association mapping designs, and local variance component analysis.
• Bayesian local variance component analysis: Introduces a novel Bayesian approach for local variance component analysis to estimate local genomic contributions to phenotypic variance.
• Scalability and performance: Demonstrates comparable analytical performance to in-memory methods on real data while providing improved memory efficiency.

Scientific Applications:

• Genome-wide association studies (GWAS): Enables GWAS on high-throughput sequencing data with datasets larger than available RAM.
• Rare variant discovery: Supports detection and testing of associations involving rare genetic variants from sequencing data.
• Nested association mapping and variance decomposition: Facilitates analysis of nested association mapping designs and local variance component analyses to attribute phenotypic variation to genomic regions.

Methodology:

Uses an HDF5-based out-of-core implementation that stores genotype data on disk and provides analytical methods including linear models, linear mixed models, rare variant testing, nested association mapping analysis, local variance component analysis, and a Bayesian local variance component approach.