Unveiling the landscape of prokaryotic global regulators through deep protein language models.

ABSTRACT: Global regulators (GRs) are key transcription factors that orchestrate the expression of multiple genes, playing essential roles in stress responses, virulence, secondary metabolism, and antibiotic resistance-traits that make them powerful tools for synthetic biology applications. However, conventional approaches often fail to detect remote homologs and novel GR types, limiting our understanding of their regulatory diversity and evolutionary dynamics across prokaryotes. Here, we present a large-scale, protein language model-driven framework to systematically chart the global regulatory landscape across 14,800 bacterial and archaeal type strain genomes-the most taxonomically diverse prokaryotic data set analyzed to date. Using a deep learning-based GR identification model trained on 74,872

SUBMITTER: Geng J 

PROVIDER: S-EPMC12710356 | biostudies-literature | 2025 Dec

REPOSITORIES: biostudies-literature