ASTAR’s HERRO sharpens nanopore sequencing, tackling error‑prone reads

HERRO corrects nanopore sequencing errors and supports single-platform workflows.

The Agency for Science, Technology and Research (A*STAR) Genome Institute of Singapore (A*STAR GIS) has developed an artificial intelligence (AI) tool that improves the accuracy of long-read genome sequencing data.

In a media release dated 2 June, A*STAR said the tool, called HERRO, corrects errors in nanopore sequencing reads produced by Oxford Nanopore Technologies (ONT), a platform used for long-read DNA sequencing to study complex regions of the genome.

HERRO uses deep learning to improve the accuracy of ONT Simplex reads—single-strand DNA reads that typically have higher error rates—by up to 100-fold.

The improvement allows researchers to generate high-quality genome assemblies using a single sequencing platform, without relying on multiple technologies.

The tool is designed to be haplotype-aware, meaning it preserves differences between chromosome copies while correcting sequencing errors, according to A*STAR GIS.

A*STAR said HERRO-enabled reads allowed researchers to reconstruct complete human chromosomes end-to-end, including telomere-to-telomere (T2T) assemblies of the X and Y chromosomes.

The tool was tested across multiple human and non-human genomes, producing results comparable to or better than multi-platform sequencing workflows, whilst requiring less DNA material.

The research team said the tool could support applications in genome mapping, genetic variation analysis, and research in disease, agriculture and biotechnology.