GeneImp: Fast Imputation to Large Reference Panels Using Genotype Likelihoods from Ultralow Coverage Sequencing.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genotype imputation is crucial for analyzing genetic data, especially from ultralow coverage sequencing where data is sparse and uncertain.
• Existing imputation methods often rely on prephasing, which is computationally expensive and challenging with uncertain genotype calls from low-coverage data.

Purpose of the Study:

• To develop a computationally efficient genotype imputation program, GeneImp, suitable for ultralow coverage sequencing data.
• To enable whole-genome imputation to dense reference panels without the need for prephasing.

Main Methods:

• GeneImp utilizes large reference panels and assumes short haplotype regions are unaltered, bypassing the need for explicit recombination modeling or prephasing.
• The program processes genotype likelihoods directly, accommodating the probabilistic nature of ultralow coverage sequencing data.

Main Results:

• GeneImp achieves imputation quality comparable to state-of-the-art methods like BEAGLE.
• GeneImp demonstrates a significant computational advantage, operating one to two orders of magnitude faster than BEAGLE.
• The program maintains similar memory complexity compared to existing methods.

Conclusions:

• GeneImp provides a practical and efficient solution for whole-genome genotype imputation from ultralow coverage sequencing data.
• Its ability to impute without prephasing makes it ideal for datasets with high missingness or uncertainty.
• Future applications include imputation using off-target reads from whole-exome sequencing.