The paper introduces PlasMAAG, a novel computational tool designed to reconstruct plasmids and cellular genomes from complex metagenomic data. Unlike traditional methods that analyze samples individually, this framework utilizes assembly–alignment graphs (AAGs) to combine genetic information across multiple samples. By integrating contrastive learning and deep-learning-based binning, the tool effectively overcomes common obstacles such as fragmented data and genetic complexity. Benchmark testing demonstrates that PlasMAAG significantly outperforms existing technologies, recovering a higher volume of near-complete plasmids while accurately identifying their bacterial hosts. Furthermore, the researchers successfully applied this method to hospital sewage samples to track antibiotic resistance and study intraplasmid diversity. Ultimately, this software provides a more comprehensive way to explore horizontal gene transfer and microbial community dynamics in diverse environments.

References:

• Líndez P P, Danielsen L S, Kovačić I, et al. Accurate plasmid reconstruction from metagenomics data using assembly-alignment graphs and contrastive learning[J]. bioRxiv, 2025: 2025.02. 26.640269.