Long-read reconstruction of many diverse haplotypes with devider

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate haplotype reconstruction is crucial for analyzing mixtures of similar DNA sequences.
• Long-read sequencing offers advantages for resolving distant alleles but requires robust error handling.
• Existing methods may struggle with the complexity and error rates of long-read data for small sequences.

Purpose of the Study:

• To develop and evaluate 'devider', a novel algorithm for haplotyping small sequences using long-read data.
• To assess devider's performance in terms of accuracy, speed, and memory usage compared to existing methods.
• To apply devider to investigate within-host diversity and evolutionary signals in microbial communities and resistance genes.

Main Methods:

• Utilizes a positional de Bruijn graph approach with sequence-to-graph alignment.
• Employs an alphabet of informative alleles for efficient processing.
• Designed as a fast, assembly-inspired method compatible with Oxford Nanopore Technologies (ONT) and Pacific Biosciences (PacBio) sequencing.

Main Results:

• On synthetic HIV data, devider recovered 97% of haplotype content with accurate abundance estimates, using minimal resources (<4 min, 1 GB RAM).
• For antimicrobial resistance (AMR) genes, devider achieved 83% haplotype recovery, significantly outperforming other methods.
• Successfully disentangled complex bacterial communities and HIV-1 coinfections on real-world PacBio and ONT datasets, and identified numerous AMR gene haplotypes with recombination blocks.

Conclusions:

• Devider provides a fast and accurate solution for haplotyping small sequences from long-read data.
• The algorithm excels in disentangling complex mixtures and estimating haplotype abundance.
• Devider is capable of uncovering evolutionary insights, such as recombination patterns, within heterogeneous microbial populations.