ERStruct

ERStruct infers population structure from large-scale, multi-ethnic whole genome sequencing (WGS) data to enable analysis of genetic diversity and population stratification.

Key Features:

• Parallel computing and GPU acceleration: Implements optimized matrix operations using parallel processing and GPU acceleration to speed computations on large-scale datasets.
• Adaptive data splitting: Splits input data adaptively to accommodate GPUs with limited memory while enabling processing of extensive genomic datasets.
• Estimation of informative principal components: Estimates the number of top principal components that are most informative for capturing population structure.
• PCA tailored for ultra-dimensional WGS data: Adapts principal component analysis to handle ultra-dimensional whole genome sequencing data and complex linkage disequilibrium patterns.

Scientific Applications:

• Population genetics: Infers population structure and genetic diversity within and between populations from WGS data.
• Human ancestry analysis: Supports inference of ancestry and relationships among multi-ethnic populations using genomic principal components.
• Disease association mapping: Provides population structure estimates for controlling stratification in genome-wide association studies (GWAS).
• Personalized medicine: Informs interpretation of genetic variation across ethnic groups relevant to precision medicine.

Methodology:

Uses principal component analysis (PCA) adapted for ultra-dimensional genomic data, estimates the number of informative principal components, and employs optimized matrix operations via parallel processing and GPU acceleration with adaptive data splitting to accommodate limited GPU memory.