Prioritizing genetic variants in GWAS with lasso using permutation-assisted tuning

Area of Science:

• Genetics and Genomics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) identify genetic variants linked to complex traits and disorders.
• Traditional single-nucleotide polymorphism (SNP) analysis in GWAS is limited in elucidating genetic architecture and can be underpowered.
• Multiple-SNP analyses offer increased power but lack the ability to identify specific causal SNPs within a set.

Purpose of the Study:

• To propose a novel permutation-assisted tuning procedure in lasso (plasso) for identifying phenotype-associated SNPs in joint multiple-SNP regression models within GWAS.
• To enhance the biological interpretability and statistical power of GWAS by pinpointing individual SNPs responsible for genotype-phenotype associations.

Main Methods:

• Developed a permutation-assisted tuning procedure (plasso) for the least absolute shrinkage and selection operator (lasso) regression model.
• Generated pseudo-single-nucleotide polymorphisms (SNPs) through permutation to serve as non-informative controls.
• Optimized the lasso tuning parameter to effectively distinguish true signal SNPs from non-informative pseudo-SNPs.

Main Results:

• Simulations demonstrated that plasso outperforms existing methods in identifying phenotype-associated SNPs.
• Application of plasso to a real GWAS dataset provided new insights into the genetic control of complex traits.
• The method successfully identifies specific SNPs driving joint genotype-phenotype associations within SNP sets.

Conclusions:

• Plasso offers a statistically robust and biologically interpretable approach for fine-mapping genetic associations in GWAS.
• The method enhances the ability to dissect the genetic architecture of complex traits by identifying key individual SNPs.
• The proposed methodology provides a valuable tool for advancing genetic research in complex diseases.