HINGE: long-read assembly achieves optimal repeat resolution.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long-read sequencing offers potential for high-quality de novo genome assembly.
• Resolving repetitive regions in genomes using error-prone long reads remains a significant challenge.
• Existing assemblers either misassemble repeats or fragment genomes.

Purpose of the Study:

• To develop a novel genome assembler, HINGE, for optimal repeat resolution in long-read sequencing data.
• To distinguish between resolvable and unresolvable repeats to improve assembly accuracy.
• To provide a robust method for handling ambiguous genomic regions.

Main Methods:

• HINGE employs a unique approach by adding "hinges" to reads to construct an overlap graph.
• This method merges only unresolvable repeats, preserving information about complex genomic structures.
• Combines error resilience of overlap-based assemblers with repeat-resolution of de Bruijn graph assemblers.

Main Results:

• HINGE produced more finished genome assemblies compared to Miniasm, HGAP, and Circlator on NCTC bacterial datasets.
• Identified 40 datasets where unresolvable repeats precluded unique finished assemblies.
• HINGE generated interpretable assembly graphs for ambiguous regions, unlike other methods that fragmented or arbitrarily resolved them.

Conclusions:

• HINGE offers an advanced solution for de novo genome assembly using long-read data.
• The assembler effectively handles repeats, leading to more complete and accurate genome reconstructions.
• HINGE provides valuable insights into genomic complexity by visualizing unresolvable repeats.