Pandoomain, a scalable pipeline for genomic and protein domain context analysis, reveals widespread PT-TG domain architectural diversity and novel polymorphic toxins

The rapid expansion of bacterial genome databases presents significant opportunities for functional discovery, as a large fraction of genes and protein domains remain uncharacterized. Analyzing genomic context and domain architecture is a powerful approach for functional inference, but existing tools often lack the scalability and integrated workflow required for high-throughput analysis. To address this, we developed Pandoomain, a Snakemake pipeline that automates the acquisition of genomes from the National Center for Biotechnology Information, identifies proteins of interest using hidden Markov models (HMMs), and performs systematic domain annotation and gene neighborhood analysis. We demonstrate the utility of Pandoomain through a comprehensive analysis of the poorly characterized pre-toxin TG (PT-TG) domain across 347,289 bacterial genomes. Our analysis revealed 10,226 PT-TG-containing proteins organized into 312 unique domain architectures, highlighting their association with diverse interbacterial antagonistic systems, including the Type VI secretion, Type VII secretion, and contact-dependent inhibition systems. By leveraging genomic context, we identified a novel variant of the WXG trafficking domain, termed W10XG, and subsequently discovered 24 new families of associated toxin domains. We experimentally validated six of these toxins, confirming that all six are neutralized by their cognate immunity proteins. Pandoomain is an accessible tool that enables systematic, large-scale exploration of protein domains, and our analysis of the PT-TG domain provides a rich resource for future investigations into the mechanisms and evolution of bacterial antagonism.IMPORTANCEThe rapid growth of bacterial genomic data presents a major hurdle for scientists seeking to understand the functions of newly discovered genes and proteins. To address this issue, we created Pandoomain, a powerful, accessible software tool that automates large-scale analysis of genetic information across hundreds of thousands of genomes. Using Pandoomain, we investigated a poorly understood family of proteins involved in bacterial competition, revealing novel protein domain architectural diversity. This led to the discovery of 24 new families of toxins predicted to be used by bacteria to attack their competitors, and we experimentally confirmed the toxic activity of six of them. Our work provides the scientific community with a robust tool to accelerate functional discovery and offers new insights into the evolution of bacterial conflicts, which may provide insights into the compositional dynamics of microbial communities and support methods to engineer their composition.