SIMreg

SIMreg performs similarity-based regression to assess associations between genetic marker sets and quantitative traits, quantifying main and interaction effects while accommodating both common and rare variants through allele-frequency–adaptive weighting.

Key Features:

• Similarity-Based Regression Method: Aggregates information across multiple polymorphic sites by computing genetic similarities and relating them to trait similarities.
• Adaptive Allele-Frequency Weighting: Integrates adaptive weights based on allele frequencies to accommodate common and rare variants without dichotomizing allele types.
• Signal Preservation via Similarity-Level Collapsing: Collapses information at the similarity level rather than the genotype level to avoid cancelation of signals with opposing etiological effects.
• Versatility Across Variant Types: Applies to any class of genetic variants, allowing analysis across diverse marker sets.
• Pairwise Regression on Unrelated Individuals: Regresses trait similarities between pairs of unrelated individuals on their genetic similarities to assess gene–trait associations.
• Statistical Testing Using Score Test: Employs a score test with a derived limiting distribution to evaluate association significance.
• Inclusion of Covariates and Interaction Effects: Supports inclusion of covariates and explicit modeling of both main and interaction effects.
• Computational Efficiency: Implemented to be computationally feasible for large-scale genomic analyses.

Scientific Applications:

• Whole-Genome and Region-Level Association Studies: Evaluates associations where marker sets are defined by linkage disequilibrium (LD) blocks, genes, or pathways.
• Detection of Modest and Complex Genetic Effects: Detects modest etiological effects and complex interaction effects among markers on quantitative traits.
• Phenotype–Marker Set Association Evaluation: Assesses associations between phenotypes and diverse marker sets to investigate the genetic architecture underlying complex traits.

Methodology:

Compute genetic similarities across polymorphic sites with allele-frequency–adaptive weights, collapse information at the similarity level, regress pairwise trait similarities of unrelated individuals on genetic similarities, and evaluate significance using a score test with a derived limiting distribution while allowing covariates and interaction terms.