MixSIH

MixSIH infers two haplotypes of a single individual (single individual haplotyping, SIH) from aligned sequence fragments using a probabilistic two-component mixture model and assigns segment-level reliability scores for haplotype assembly.

Key Features:

• Probabilistic two-component mixture model: Represents the two haplotypes present in an individual's genomic data with two mixture components to drive assembly.
• Minimum connectivity (MC) score: Computes an MC score for each assembled segment to indicate segment reliability and accuracy.
• Pairwise consistency assessment: Uses pairwise consistency measures to evaluate the accuracy of assembled haplotypes.
• Chimeric read fragment handling: Identifies and mitigates chimeric read fragments derived from different haplotypes to reduce assembly errors.
• Extraction of high-accuracy segments: Produces highly accurate partial haplotype segments from aligned fragment data.
• C++ implementation: Provided as open-source C++ code.

Scientific Applications:

• Haplotype assembly for GWAS: Supplies phased haplotype information useful for genome-wide association studies and other genetic analyses.
• Segment reliability for downstream analyses: Uses MC scores to prioritize reliable haplotype segments for downstream genetic analysis.
• Benchmarking and evaluation: Enables accuracy evaluation of haplotype assemblies on simulated and real sequencing datasets via pairwise consistency.
• Chimera detection in sequencing data: Detects chimeric read fragments that can degrade haplotype assembly quality.

Methodology:

Uses a probabilistic model with two mixture components to represent dual haplotypes; computes a minimum connectivity (MC) score per segment; applies pairwise consistency measures to evaluate assembled haplotypes; identifies and mitigates chimeric read fragments; implemented in C++ and evaluated on simulated and real datasets.