SNPInt-GPU

SNPInt-GPU performs statistical epistasis testing to detect interactions between single nucleotide polymorphisms (SNPs) and assess their combined effects on phenotypic traits.

Key Features:

• GPU Acceleration: Uses the Nvidia CUDA framework for GPU acceleration to increase computational efficiency on large-scale genomic datasets.
• Diverse Statistical Methods: Implements logistic regression (as in PLINK epistasis testing), BOOST (Bayesian Optimization of SNP Interactions), log-linear regression, and mutual information (MI) and information gain (IG) for pairwise and third-order tests.
• Linkage Disequilibrium Analysis: Optionally calculates r^2 scores on-the-fly to assess linkage disequilibrium between loci.

Scientific Applications:

• Epistasis analysis: Tests SNP–SNP interactions to evaluate their combined influence on binary and other phenotypic traits.
• Genetic marker discovery: Identifies interacting SNPs that may serve as candidate genetic markers and aid in understanding mechanisms of multifactorial and hereditary conditions.

Methodology:

Implements GPU acceleration via Nvidia CUDA; performs logistic regression (PLINK-like), BOOST (Bayesian Optimization of SNP Interactions), log-linear regression, mutual information (MI) and information gain (IG) for pairwise and third-order epistasis tests; optionally computes r^2 scores on-the-fly for LD; developed for Linux-based systems and uses CUDA libraries for optimal performance.