Improving population scale statistical phasing with whole-genome sequencing data

Area of Science:

• Genomics
• Population Genetics
• Bioinformatics

Background:

• Haplotype estimation (phasing) is crucial for population genetics and variant analysis.
• Statistical phasing methods scale well but struggle with rare variants due to limited data.
• Existing methods often disregard valuable read-level sequencing information.

Purpose of the Study:

• Introduce SAPPHIRE, a novel method to enhance haplotype call precision.
• Leverage whole-genome sequencing data to improve accuracy of statistical phasing.
• Refine low-confidence phase calls using read-level information.

Main Methods:

• Developed SAPPHIRE, a method integrating whole-genome sequencing reads.
• Refined haplotype estimates by realigning sequencing reads, focusing on low-confidence calls.
• Applied SAPPHIRE to large-scale datasets, including UK Biobank data.

Main Results:

• SAPPHIRE significantly improves the accuracy of state-of-the-art phasing methods.
• Identified a subset of phase calls validated by direct sequencing read evidence.
• Demonstrated scalability of SAPPHIRE on 3.6 Petabytes of UK Biobank data.

Conclusions:

• SAPPHIRE enhances haplotype estimation accuracy by incorporating sequencing read data.
• The method effectively refines low-confidence calls and validates phasing results.
• SAPPHIRE is a scalable tool applicable to massive genomic datasets for improved variant analysis.