GWAS-Flow

GWAS-Flow implements GPU-accelerated genome-wide association studies using TensorFlow to compute linear mixed models and permutation-based significance thresholds for large-scale genotyping and phenotyping data.

Key Features:

• Permutation-Based Significance Thresholds: Implements permutation-based significance thresholds that adjust for multiple hypothesis testing and account for non-Gaussian phenotypic distributions.
• Linear Mixed Model Implementation: Implements a linear mixed model (LMM) framework for association testing of complex traits with genomic polymorphisms.
• GPU Acceleration: Utilizes GPU infrastructure via TensorFlow to reduce computation time without compromising accuracy.
• Versatile Computational Options: Provides both CPU and GPU execution modes, with the CPU mode for smaller datasets and the GPU mode for large datasets.
• Comprehensive Output Metrics: Reports p-values, effect sizes, and standard errors for each genetic marker.

Scientific Applications:

• Large-Scale GWAS: Analyzes large genotyping and phenotyping datasets for genome-wide association studies.
• Non-Gaussian Phenotype Analysis: Produces significance estimates robust to non-Gaussian phenotypic distributions via permutation testing.
• Complex Trait Association Mapping: Identifies associations between genomic polymorphisms and complex traits.

Methodology:

Implements a linear mixed model developed with TensorFlow and computes permutation-based significance thresholds, leveraging GPU acceleration for computational performance and producing p-values, effect sizes, and standard errors.