Chimeric reference panels for genomic imputation.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Missing genotype data is a significant limitation in genomic research.
• Current genotype imputation methods require expensive external reference panels, limiting their use in nonmodel organisms.
• Existing methods often fail to capture novel genomic variations not present in reference panels.

Purpose of the Study:

• To develop a novel method, Retriever, for constructing chimeric reference panels for genotype imputation.
• To enable accurate genotype imputation without relying on external reference panels.
• To provide a cost-effective and accessible solution for genomic analyses across diverse species.

Main Methods:

• Retriever constructs a chimeric reference panel directly from target samples.
• A sliding window approach identifies and retrieves genomic partitions with complete data.
• Exploits complementary missing data distributions across samples to assemble a reference panel.

Main Results:

• Retriever successfully constructs reference panels that preserve local linkage disequilibrium and capture novel variants.
• Genotype imputation using Retriever-constructed panels with Beagle consistently achieves accuracy exceeding 95%.
• The method demonstrates high accuracy across diverse datasets, including plants, animals, and fungi.

Conclusions:

• Retriever eliminates the need for costly external reference panels for genotype imputation.
• This method offers an accessible and cost-effective solution, broadening the scope of genomic research.
• Retriever enhances the potential of genomic analyses by addressing the challenge of missing data.