TinkerHap

TinkerHap performs haplotype phasing of sequencing data by linking reads across heterozygous sites to assign alleles to parental chromosomes and improve variant-level phasing accuracy.

Key Features:

• Read-based phaser: Links sequencing reads across heterozygous sites to phase alleles and extend haplotype blocks.
• Pairwise distance-based unsupervised classification: Uses a pairwise distance–based unsupervised classification system to group reads for haplotype construction.
• Hybrid integration: Incorporates external pre-phased data from statistical or pedigree sources to augment read-based phasing.
• Sequencing platform support: Demonstrated on Illumina short-reads and PacBio long-reads.
• Benchmarking datasets: Evaluated on 1,040 UK Biobank parent-offspring trios and the GIAB Ashkenazi trio.
• Accuracy metrics: Read-based phasing accuracies reported as 95.1% for short reads and 97.5% for long reads, with a hybrid short-read accuracy of 96.3% and 99.5% of heterozygous sites phased in the hybrid approach.
• Haplotype block extension: Achieves a median long-read haplotype block size of 79,449 bp (compared to a second-best 68,303 bp).
• SNP and indel phasing: Reports higher phasing accuracy for both single nucleotide polymorphisms (SNPs) and insertions/deletions (indels).

Scientific Applications:

• Genetic variation analysis: Phasing haplotypes to study genetic variation within cohorts.
• Disease-variant identification: Assigning alleles to parental chromosomes to aid identification of disease-associated variants.
• Phasing benchmarking: Comparing phasing performance across sequencing technologies and methods using trio datasets such as UK Biobank and GIAB.

Methodology:

Combines a read-based phaser that links sequencing reads across heterozygous sites with a pairwise distance-based unsupervised classification system and optionally integrates pre-phased statistical or pedigree data in a hybrid approach.