FPGA_perm

FPGA_perm accelerates permutation testing for genome-wide association studies (GWAS) by implementing maxT and adaptive permutation testing on Field-Programmable Gate Arrays (FPGAs) to reduce runtime on large-scale SNP datasets.

Key Features:

• FPGA acceleration: Implements permutation testing on Field-Programmable Gate Arrays (FPGAs) to enable large-scale parallel computation and reduced runtimes.
• Cloud FPGA compatibility: Optimized for AWS EC2 F1 instances (e.g., f1.2xlarge) to leverage cloud-based FPGA resources for scalable computation.
• Permutation algorithms: Supports maxT permutation testing and adaptive permutation testing for controlling false positive rates in GWAS.

Scientific Applications:

• GWAS permutation testing: Enables permutation-based significance assessment and false discovery control in large-scale GWAS analyzing millions of single nucleotide polymorphisms (SNPs) across thousands of individuals.

Methodology:

Implements maxT and adaptive permutation testing on FPGA hardware using parallelization on FPGAs; benchmarked against CPU-based implementations (PLINK) on a system with 40 Intel Xeon 4114 CPU cores using an imputed breast cancer dataset with 13.7 million SNPs from 3,652 individuals, producing runtimes of ~22 minutes for 1,000 maxT permutations (versus nearly seven days on 40 Intel Xeon 4114 cores), ~325 minutes for 100 million adaptive permutations (versus ~8.5 days for PLINK), and ~33 hours for 700 million adaptive permutations (versus an almost unfeasible runtime on a traditional multi-core CPU setup).