UNPHASED

UNPHASED performs maximum-likelihood genetic association analysis of haplotypes and genotypes in nuclear families and unrelated subjects while accommodating missing data, uncertain haplotype phase, and robustness to population stratification.

Key Features:

• Maximum-likelihood inference: Uses maximum-likelihood inference to model haplotype and genotype effects.
• Robust likelihood model for nuclear families: Implements a likelihood model with distinct association parameters for parental and offspring genotypes to enhance robustness to population stratification and missing data.
• Handling missing data and uncertain haplotype phase: Manages missing family members and uncertain haplotype phase and represents unrelated subjects as offspring of two hypothetical parents to integrate them into analyses.
• Conditioning step for multiple offspring: Incorporates a conditioning step to ensure valid analyses when multiple offspring are present, particularly in the presence of linkage.
• Comparison with existing tools: Compared to TRANSMIT it eliminates bias from missing data in the presence of linkage; it is slightly less robust to population structure than FBAT and PCPH but shows greater power for strong genetic effects; relative to APL and MITDT it demonstrates enhanced robustness to stratification and accommodates sibships of any size.
• Versatility for trait analysis: Applicable to both binary and continuous (quantitative) traits.

Scientific Applications:

• Family-based association studies: Association analysis in nuclear families and family-based designs accounting for missing data and phase uncertainty.
• Integration of unrelated subjects: Joint analysis of unrelated subjects and family data by modeling unrelateds as offspring of hypothetical parents.
• Linkage- and stratification-aware mapping: Detection of genetic associations while accounting for linkage and population stratification.
• Binary and quantitative trait mapping: Mapping genetic effects for both disease (binary) and quantitative traits.

Methodology:

Performs maximum-likelihood inference with distinct association parameters for parental and offspring genotypes, represents unrelated subjects as offspring of two hypothetical parents, and applies a conditioning step for analyses with multiple offspring.