Omni-ARMD: an integrated genomic and clinical knowledgebase bridging clinical phenotypes and molecular determinants of antimicrobial resistance

Abstract

Antimicrobial resistance (AMR) is a major global health threat, yet its molecular determinants remain difficult to characterize in real-world settings, as genomic and clinical data are fragmented across resources. Existing databases typically separate gene-level mechanisms, curated genotype–phenotype associations, and large-scale clinical antimicrobial susceptibility testing (AST) data, limiting cross-scale interpretation. Here, we present Omni-ARMD, an open-access knowledgebase that integrates multi-source genomic and clinical data to systematically infer genotype–phenotype relationships in AMR. Omni-ARMD bridges large-scale clinical AST records and gene-level resistance evidence through a confidence-weighted log-odds enrichment framework, which quantifies genotype–phenotype associations at the organism–antimicrobial level and integrates these associations into aggregated cross-domain evidence representations. The current release includes over 12 million de-identified microbiology records from more than 1 million patients, harmonized to generate over 7.83 million standardized AST records. It also integrates over 0.36 million gene-phenotype associations, spanning 38 organisms, 52 antimicrobials, and 375 resistance genes. Omni-ARMD links over 5.16 million clinical records to genomic evidence, achieving a 65.86% match rate and a 2.61% coverage of organism–antimicrobial pairs. Omni-ARMD provides interactive modules for clinical investigation, genomic evidence analysis, mechanistic interpretation, and artificial intelligence-assisted exploration, enabling integrative and data-driven discovery in AMR. Omni-ARMD is freely available at www.omniarmd.cn.