A Genome-Based Pipeline for Digital Siderophore Typing in Pseudomonas Iron-Interaction Network

Siderophores are key mediators of microbial interactions, especially in iron-limited environments. While traditional experimental methods for siderophore typing are time-consuming and low-throughput, we present a genome-based pipeline for high-throughput digital siderophore typing in Pseudomonas. Our approach integrates profile Hidden Markov Models (pHMMs), substrate-specific motif identification, and co-evolutionary analysis to predict both the structure of pyoverdines and their corresponding uptake receptors from any given Pseudomonas genome. By this approach, we developed pHMM from 94 previously defined receptor groups and validated their accuracy and robustness across 14,230 Pseudomonas genomes. Application of our method to this large dataset demonstrated the potential of our algorithm for identifying both novel "lock-key" receptor groups and previously uncharacterized pyoverdine structures. Notably, our pipeline corrected long-term misclassifications in classical strains and proposed a new reference for the canonical Group III pyoverdine. Furthermore, interaction network analysis supports the observation of distinct siderophore utilization patterns between pathogenic and non-pathogenic strains. This standardized, user-friendly platform offers a robust tool for annotating siderophore behaviors in Pseudomonas and demonstrates the potential of digital siderophore typing in exploring iron-mediated ecology across microbes. Impact statementIron is vital for microbes. In most environments, they secrete diverse siderophores to scavenge iron. The types of siderophores a strain produces and uptakes shape its interactions, from cooperation to cheating and competition; characterizing this is called "siderophore typing." While traditional methods are limited by experimental capacities, our genome-based digital pipeline revolutionizes the process by predicting siderophore structures and receptor types directly from genomes, enabling large-scale analysis of 14,230 Pseudomonas strains. The results reveal novel groups, correct historical misclassifications, and link patterns to pathogenicity, providing a standardized software for ecological studies, antibiotic design, and bioremediation.