Short communication: imputing genotypes using PedImpute fast algorithm combining pedigree and population information.

Area of Science:

• Animal Genetics
• Bioinformatics
• Genomic Selection

Background:

• Genomic evaluations in livestock rely on accurate imputation of missing genotypes.
• Existing imputation algorithms face challenges with computational efficiency and accuracy, particularly for routine applications.

Purpose of the Study:

• To develop and evaluate PedImpute, a novel algorithm for genomic imputation.
• To compare PedImpute's performance against findhap and BEAGLE using Holstein cattle genotype data.

Main Methods:

• Genomic data from 19,904 Holstein animals across four countries were used.
• PedImpute was compared with findhap and BEAGLE based on imputation accuracy, error rates, and computational time.
• Performance was assessed under various genotyping density scenarios, including low-density chips.

Main Results:

• PedImpute demonstrated slightly higher accuracy and faster computation than findhap and BEAGLE when parentage (sire, dam, maternal grandsire) was densely genotyped.
• BEAGLE showed better performance for animals with sparsely genotyped relatives, but its computational demands were prohibitive for routine use.
• PedImpute achieved error rates of 0.2-1.1% and allelic squared correlations of 96.6-99.3%.

Conclusions:

• PedImpute is a valuable and efficient option for routine genomic evaluations, especially when complete parentage genomic information is available.
• The algorithm balances high accuracy with reduced computing time, crucial for large-scale livestock breeding programs.
• Future availability of dense parentage genotypes in dairy cattle supports PedImpute's utility.