Fast and Robust Identity-by-Descent Inference with the Templated Positional Burrows-Wheeler Transform.

Area of Science:

• Genomics
• Computational Biology
• Population Genetics

Background:

• Accurate estimation of identical-by-descent (IBD) segments is vital for genetic analyses.
• Increasing biobank and direct-to-consumer genetic data present computational challenges for IBD inference.

Purpose of the Study:

• To introduce the templated positional Burrows-Wheeler transform (TPBWT) for efficient and accurate IBD segment estimation.
• To assess TPBWT's performance against existing IBD inference methods, particularly concerning genotype and phasing errors.

Main Methods:

• Developed and implemented the templated positional Burrows-Wheeler transform (TPBWT) algorithm.
• Simulated haplotype data with realistic genotyping and phasing errors for performance evaluation.
• Compared TPBWT with state-of-the-art phase-aware and phase-free IBD inference methods.
• Introduced in-sample and out-of-sample TPBWT algorithms and a binary file format for compressed haplotypes.

Main Results:

• TPBWT demonstrated superior speed and accuracy compared to existing IBD inference algorithms.
• TPBWT showed robustness to genotype and phasing errors, unlike many phase-aware methods.
• Computational efficiency was confirmed on massive datasets with millions of samples.
• Empirical analysis revealed geographically structured haplotype sharing and isolation by distance within Mexico.

Conclusions:

• TPBWT offers a computationally efficient and accurate solution for IBD segment estimation in large genetic datasets.
• The method's robustness to phasing errors makes it suitable for real-world genetic data.
• TPBWT facilitates large-scale population genetic analyses, such as exploring geographic patterns of genetic sharing.