mrMLM

mrMLM performs multi-locus genome-wide association analysis by modelling random SNP effects under a mixed linear model to improve detection of quantitative trait loci (QTL).

Key Features:

• Implementation: mrMLM v4.0.2 is implemented in R.
• Integration of methods: Integrates mrMLM, FASTmrMLM, FASTmrEMMA, pLARmEB, pKWmEB, and ISIS EM-BLASSO for multi-locus GWAS analyses.
• Efficient data handling: Uses fread from the data.table package for fast data input and employs parallel computation via the doParallel package to utilize multiple CPUs.
• Statistical framework: Operates under a multi-locus random-SNP-effect mixed linear model with genome-wide marker scanning, marker selection using a threshold less stringent than Bonferroni correction, empirical Bayes effect estimation, and likelihood ratio tests to identify non-zero effects.
• Performance validation: Validated on real and simulated datasets and reported to control false positive rates while improving QTL detection power relative to other methods.

Scientific Applications:

• Plant and animal breeding: Applicable to field experiments and breeding programs where large phenotypic errors can obscure genetic associations, improving QTL detection for selection decisions.
• Complex trait dissection: Suited for mapping quantitative trait loci underlying complex traits by incorporating multiple loci into association models.

Methodology:

Reads input with fread (data.table), performs parallel computation via doParallel, integrates the listed multi-locus methods, scans each genome marker, selects markers with a threshold less stringent than Bonferroni, estimates marker effects using empirical Bayes, and identifies non-zero effects with likelihood ratio tests.