pLAST - a tool for rapid comparison and classification of bacterial plasmid sequences

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• The rapid growth of sequenced bacterial plasmids necessitates advanced computational tools for comparison and classification.
• Current methods often struggle with scalability or are limited to closely related plasmids, hindering large-scale analysis.

Purpose of the Study:

• To develop a novel, efficient computational tool for bacterial plasmid comparison and classification.
• To improve the identification of functionally similar plasmids and shared genetic elements.

Main Methods:

• Developed pLAST (plasmid Language Analysis and Search Tool), utilizing word2vec representations of protein-family content within local genomic context.
• Benchmarked pLAST against established nucleotide-based (e.g., Mash) and ORF-based (e.g., DIAMOND) methods.

Main Results:

• pLAST demonstrates superior performance over nucleotide-based methods and comparable results to DIAMOND in identifying functionally similar plasmids.
• Achieved significant improvements in detecting shared mating-pair formation (MPF) system types (26%) and relaxase types (24%) compared to Mash.
• Successfully scaled to analyze large datasets, including a collection of ~750,000 plasmids, and provides per-ORF alignments for module detection.

Conclusions:

• pLAST offers a computationally efficient and accurate solution for large-scale bacterial plasmid analysis.
• The tool enhances the detection of shared functional modules and key plasmid features, advancing plasmid research.