A deep learning-based method enables the automatic and accurate assembly of chromosome-level genomes

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• High-throughput chromosome conformation capture (Hi-C) technology is crucial for chromosome-level genome assemblies.
• Challenges persist in error correction and sequence anchoring within these assemblies.

Purpose of the Study:

• To develop an automated deep learning-based method, AutoHiC, for enhancing genome assembly contiguity and accuracy.
• To address limitations of manual refinement in conventional Hi-C-aided scaffolding.

Main Methods:

• Developed AutoHiC, a deep learning model utilizing Hi-C data for automated workflows.
• Implemented iterative error correction within the AutoHiC framework.
• Trained AutoHiC on Hi-C data from over 300 species.

Main Results:

• AutoHiC demonstrated an average error detection accuracy exceeding 90% across diverse species.
• Benchmarking confirmed significant improvements in genome contiguity and error correction.
• The method automates processes typically requiring manual intervention.

Conclusions:

• AutoHiC represents a breakthrough in automated error detection for genome assembly.
• The method promises more accurate and contiguous chromosome-level genome assemblies.
• This advancement will significantly benefit future genomics research by providing reliable genome data.