OCMA

Out-of-Core Matrices Analyzer (OCMA) is a disk-based computational tool to attack the challenges posed by the analysis of large genomic relationship matrices (GRMs) and genotype matrices that are central to genetic analysis but too voluminous to fit into the active memory of standard computing resources. With the advent of large-scale biobanks, the size of these matrices has outpaced the memory capabilities of conventional systems, necessitating innovative solutions to facilitate essential genetic analyses such as SNP-heritability estimation, Principal Component Analysis (PCA), and genomic prediction.

OCMA addresses this challenge by employing state-of-the-art computational techniques that allow for efficient eigen and Singular Value Decomposition (SVD) analyses without needing the data to reside entirely in memory. It utilizes memory mapping (mmap) techniques alongside the latest matrix factorization libraries to achieve high-speed, memory-efficient operations, making OCMA an invaluable tool for genetic researchers dealing with "bigger-data" scenarios, enabling them to perform complex analyses on standard personal computers or high-performance computing (HPC) clusters.

The performance of OCMA is notable, demonstrating its capability to perform full eigen-decomposition of a standard GRM with 10,000 individuals in just 55 seconds on a personal computer. For SVD—a faster alternative for many genomic analyses—OCMA can compute the top 200 singular values in just half an hour, the top 2,000 singular values in approximately 0.95 hours, and all 5,000 singular values in 1.77 hours for a genotype matrix of 1,000,000 individuals by 5,000 markers.