BIMBAM

BIMBAM performs genotype imputation and Bayesian association testing to evaluate single nucleotide polymorphisms (SNPs) and regional multi-SNP associations in genetic studies.

Key Features:

• Imputation-Based Association Methods: Assesses untyped variants via imputation to test associations with phenotypes for single SNPs and regional multi-SNPs.
• Factors Affecting Imputation Accuracy: Accounts for practical influences on imputation accuracy, including the choice of reference panel, and how these factors affect power to detect associations.
• Bayesian vs. Frequentist Approaches: Supports evaluation and comparison of Bayesian and frequentist tests and notes that ranking SNPs by a likelihood ratio test can be equivalent to a Bayesian procedure under specific prior assumptions.
• Computational Efficiency: Approximates full Bayesian analyses by replacing unknown genotypes with their posterior mean to reduce computational demands and enable genome-wide analyses with modest resources.
• Facilitating Multi-Study Integration: Combines information across studies using only summary data for each SNP to support meta-analysis and cross-platform integration.

Scientific Applications:

• Genome-wide association studies (GWAS): Detects associations between genetic variants and phenotypes at genome-wide scale using imputed and typed genotypes.
• Meta-analysis and multi-study integration: Integrates summary SNP-level data across studies and genotyping platforms to increase power.
• Power and accuracy assessment: Evaluates the impact of imputation accuracy and reference panel choice on the power to detect associations.

Methodology:

Uses Bayesian statistical methods to impute untyped variants and test single SNP and regional multi-SNP associations, approximates full Bayesian analyses by replacing unknown genotypes with their posterior mean, compares Bayesian and frequentist tests (noting likelihood ratio test equivalence under specific priors), and combines study information using SNP-level summary data.