Kinpute: using identity by descent to improve genotype imputation

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Genotype imputation accuracy is limited by reference panel representation.
• Individuals sharing genomic regions identical by descent (IBD) can improve imputation.
• Existing methods struggle with poorly imputed genotypes in specific populations.

Purpose of the Study:

• To develop a novel method for improving genotype imputation accuracy.
• To leverage IBD information and study-specific reference panels (SSRPs) for enhanced imputation.
• To provide a software tool that works even without pedigree information.

Main Methods:

• Kinpute utilizes IBD information (familial or ancestral) and linkage disequilibrium (LD) imputation outputs.
• A novel algorithm selects an optimal SSRP from study samples.
• Combines LD-imputed genotype probabilities with IBD configurations for improved accuracy.

Main Results:

• Kinpute significantly improves genotype imputation accuracy, particularly for low-certainty imputations.
• Demonstrated substantial improvement on a human population isolate dataset.
• Achieved higher accuracy genotype probabilities compared to standard LD-based imputation (Impute2).

Conclusions:

• Kinpute offers a substantial improvement in genotype imputation quality.
• The software is effective even in the absence of a known pedigree.
• Kinpute enhances the utility of genetic studies by improving genotype accuracy.