A new approach for efficient genotype imputation using information from relatives.

Area of Science:

• Animal Genetics
• Bioinformatics
• Computational Biology

Background:

• Genotype imputation reduces costs in genomic selection by inferring missing genetic markers.
• Existing Hidden Markov Model-based methods face computational challenges with large populations.
• A novel deterministic approach leverages family and population data for efficient imputation.

Purpose of the Study:

• To develop a fast and accurate genotype imputation method for large populations.
• To utilize both family and population-derived haplotype information.
• To overcome computational limitations of existing imputation techniques.

Main Methods:

• A deterministic approach using overlapping sliding windows to find haplotype matches.
• Starts with family imputation (if pedigree available), then exploits broader relationships.
• Progressively shrinks window size to capture increasingly distant relationships.

Main Results:

• Achieved higher or similar imputation accuracy compared to Beagle and Impute2 in cattle datasets.
• Demonstrated superior accuracy when close relatives were in the reference group, even without pedigree.
• Imputed rare variants with higher accuracy and significantly reduced computing time (28 minutes for 2,000 individuals).

Conclusions:

• The new method efficiently uses relatedness information for accurate genotype imputation.
• Its deterministic nature ensures speed, making it practical for large datasets.
• Offers a computationally efficient alternative to existing methods for genomic selection.