LinkImpute: Fast and Accurate Genotype Imputation for Nonmodel Organisms

Area of Science:

• Genomics
• Bioinformatics
• Population Genetics

Background:

• Genotyping is crucial for genomic studies, but missing data is a common issue.
• Genotype imputation methods improve downstream analysis power but often rely on high-quality reference genomes and genetic maps.
• Nonmodel organisms typically lack these resources, necessitating specialized imputation approaches.

Purpose of the Study:

• To develop and evaluate a novel genotype imputation method for nonmodel organisms with poorly developed genomic resources.
• To address the challenge of missing data in unphased genotype datasets with unordered markers.

Main Methods:

• Introduced LinkImpute, a software package utilizing a k-nearest neighbor genotype imputation method (LD-kNNi).
• LD-kNNi is designed for unordered markers and unphased genotype data from heterozygous species.
• The method leverages genome-wide marker distribution rather than physical or genetic proximity.

Main Results:

• LD-kNNi demonstrated comparable accuracy to existing methods while being computationally fast.
• The method exhibited minimal bias in allele frequency estimates.
• Successful application was shown using genotyping-by-sequencing data from apples, grapes, and maize.

Conclusions:

• LinkImpute provides an effective solution for genotype imputation in nonmodel organisms.
• The LD-kNNi method offers a robust alternative when genetic maps are unavailable or unreliable.
• This tool enhances the feasibility of large-scale genomic studies in diverse, under-resourced species.