FAPI: Fast and accurate P-value Imputation for genome-wide association study

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genotype imputation is standard in genome-wide association studies (GWAS) for analyzing untyped common variants.
• Meta-analysis of GWAS summary statistics enhances discovery but is hindered by varying variant sets, making traditional imputation impossible.
• Existing methods cannot impute summary statistics from different variant sets into a common reference panel.

Purpose of the Study:

• To develop a novel method for imputing P-values from GWAS summary statistics across different variant sets.
• To enable meta-analyses of diverse GWAS data for enhanced disease-associated loci discovery.
• To create a computationally efficient and accurate tool for P-value imputation.

Main Methods:

• Developed a P-value imputation (FAPI) method using only common variant summary statistics.
• Evaluated computational cost, showing linear complexity with untyped variants.
• Assessed accuracy against IMPUTE2 with prephasing.
• Developed a metric based on FAPI to detect abnormal variant associations.

Main Results:

• FAPI demonstrates comparable accuracy to leading genotype imputation methods like IMPUTE2.
• The method's computational cost scales linearly with the number of untyped variants.
• The novel abnormal association detection metric significantly outperforms LD-PAC in power.
• A user-friendly software tool implementing FAPI is available.

Conclusions:

• FAPI provides a fast and accurate solution for imputing GWAS summary statistics across diverse variant sets.
• This facilitates more powerful meta-analyses, leading to greater discovery of disease-associated loci.
• The developed abnormal association detection metric offers improved sensitivity for identifying spurious associations.