Tigmint: correcting assembly errors using linked reads from large molecules

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genome assembly reconstructs genomes from short DNA sequencing reads, facing challenges like errors, gaps, and repetitive sequences.
• Existing tools for misassembly correction primarily use short-read sequencing data (e.g., Illumina).
• A gap exists for tools utilizing long-range information from linked reads (e.g., 10x Genomics Chromium) for assembly refinement.

Purpose of the Study:

• To introduce Tigmint, a novel tool designed to identify and correct genome misassemblies using linked-read data.
• To evaluate Tigmint's effectiveness in improving genome assembly quality, particularly contiguity and accuracy.

Main Methods:

• Application of Tigmint to human genome assemblies generated using short reads and various assemblers (e.g., ABySS 2.0).
• Comparative analysis of assembly quality metrics (e.g., QUAST, scaffold NGA50) before and after Tigmint correction.
• Integration of Tigmint with scaffolding tools (e.g., ARCS) and evaluation using linked-read and long single-molecule sequencing data.

Main Results:

• Tigmint reduced misassemblies in an ABySS 2.0 human genome assembly by 27% (216 misassemblies).
• Scaffolding with Tigmint and ARCS improved scaffold NGA50 over five-fold, from 3 Mbp to 16.4 Mbp.
• Demonstrated Tigmint's utility in correcting assemblies from multiple tools and refining long single-molecule sequencing assemblies.

Conclusions:

• Assembly correction with Tigmint, followed by scaffolding, yields significantly more contiguous and accurate genome assemblies.
• Combining single-molecule sequencing with linked reads via Tigmint achieves high contiguity in both sequence and scaffold levels.
• Tigmint represents a significant advancement in genome assembly refinement, particularly for complex genomes.