ICGRM

ICGRM constructs genomic relationship matrices (GRMs) from genome-wide single nucleotide polymorphism (SNP) data to enable genomic prediction and genetic evaluation.

Key Features:

• Memory Efficiency: Divides genome-wide SNPs into multiple segments and computes per-segment summary statistics to substantially reduce RAM requirements.
• Scalability: Handles large datasets by segmenting SNP data; in simulations, dividing the genome into 5–200 parts reduced memory requirements from 218 GB to 14 GB.
• Integration of Summary Statistics: Aggregates per-segment summary statistics to produce a comprehensive, genome-wide GRM while maintaining matrix accuracy.
• Implementation: Implemented in C/C++ for high-performance computation.

Scientific Applications:

• Genomic prediction: Enables estimation of genetic values using genome-wide SNPs rather than pedigree data for prediction models.
• Animal breeding: Produces GRMs for genetic evaluation and selection in livestock populations.
• Human disease-risk assessment: Facilitates construction of GRMs used in studies predicting disease risk from genome-wide SNP data.

Methodology:

Segment genome-wide SNPs into parts; compute per-segment summary statistics related to GRMs with minimal RAM usage; integrate these segment-level statistics to form a complete genome-wide GRM.