Advancing long-read metagenomic binning via single-copy-gene guided contrastive learning
- 1Fudan University
Abstract
Long-read sequencing advances metagenomics by producing highly contiguous assemblies and more complete metagenome-assembled genomes (MAGs). However, current long-read metagenomic binners fail to incorporate the rich information of long-read assemblies into representation learning and exhibit limited performance on complex datasets. Here, we show that a higher proportion of long-read assembled contigs contain single-copy genes (SCGs) and more SCGs per contig. Therefore, we developed SCGBinner, which leverages SCG-guided contrastive learning to exploit the advantage of long-read data for learning high-quality contig embeddings. SCGBinner consistently outperforms other binning methods across five simulated and seven real-world long-read datasets, especially on real-world high-diversity samples. For a deep agricultural soil metagenome, SCGBinner recovered 71% more high-quality MAGs and 38% more near-complete MAGs than the second-best method. Notably, SCGBinner uniquely recovered 449 novel high-quality species, which shed light on the predicted ecological roles of 65 uncharacterised families and 93 novel genera. Overall, SCGBinner could harness the potential of long-read sequencing to provide unprecedented insights into the microbial dark matter of complex microbial communities.
Related
- Revealing pan-cancer clonal niches of T cells from single-cell RNA sequencing using contrastive learning
- Predicting miRNA–Disease Associations Based on Contrastive Learning of Linear and Nonlinear Features
- Cell-Lens: Advancing LLM Reasoning at Single-Cell Resolution
- CMVAE: Contrastive Supervision Improves Single-Omic Latent Manifold for Breast Cancer Survival Prognosis
- How effective are contrastive learning-based approaches for activity-cliff prediction?
BibTeX
@misc{han2026advancing,
title = {{Advancing long-read metagenomic binning via single-copy-gene guided contrastive learning}},
author = {H. Han and L. F. Messer and C. Quince and G. D. Bending and S. Raguideau and Z. Wang and S. Zhu},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.07.749871},
url = {https://www.biorxiv.org/content/10.64898/2026.09.07.749871v1},
note = {CC BY-ND},
}
RIS
TY - UNPB
TI - Advancing long-read metagenomic binning via single-copy-gene guided contrastive learning
AU - H. Han
AU - L. F. Messer
AU - C. Quince
AU - G. D. Bending
AU - S. Raguideau
AU - Z. Wang
AU - S. Zhu
PY - 2026
DA - 2026/09/07
PB - bioRxiv
DO - 10.64898/2026.09.07.749871
UR - https://www.biorxiv.org/content/10.64898/2026.09.07.749871v1
AB - Long-read sequencing advances metagenomics by producing highly contiguous assemblies and more complete metagenome-assembled genomes (MAGs). However, current long-read metagenomic binners fail to incorporate the rich information of long-read assemblies into representation learning and exhibit limited performance on complex datasets. Here, we show that a higher proportion of long-read assembled contigs contain single-copy genes (SCGs) and more SCGs per contig. Therefore, we developed SCGBinner, which leverages SCG-guided contrastive learning to exploit the advantage of long-read data for learning high-quality contig embeddings. SCGBinner consistently outperforms other binning methods across five simulated and seven real-world long-read datasets, especially on real-world high-diversity samples. For a deep agricultural soil metagenome, SCGBinner recovered 71% more high-quality MAGs and 38% more near-complete MAGs than the second-best method. Notably, SCGBinner uniquely recovered 449 novel high-quality species, which shed light on the predicted ecological roles of 65 uncharacterised families and 93 novel genera. Overall, SCGBinner could harness the potential of long-read sequencing to provide unprecedented insights into the microbial dark matter of complex microbial communities.
ER -