Artificial Intelligence-Integrated Proteome-Wide Target Prioritization for Drug Discovery against Multidrug-Resistant...

Neisseria gonorrhoeae causes ∼82.4 million infections annually and is rapidly approaching an untreatable status. The pathogen has developed resistance to every antibiotic class introduced since the sulfonamide era. Ceftriaxone, the last WHO- and CDC-recommended empirical therapy, is now compromised by high-level resistance in extensively drug-resistant lineages reported across multiple continents. No licensed vaccine exists, and no antibiotic with a novel mechanism has been introduced for this pathogen in over 40 years. Conventional drug discovery is too slow and insufficiently resistance-aware to address this escalating threat. In this review, we examine in silico proteome-wide target fishing as a rapid, systematic approach. The approach applies six biologically informed filters to the N. gonorrhoeae core proteome: core-genome definition, host-homology exclusion, essentiality screening, metabolic chokepoint identification, subcellular localization, and resistance-aware prioritization. Across data sets spanning 12-69 clinical strains, this pipeline reduces 2000-12,300 protein-coding sequences to 12-30 high-confidence targets. An integrated artificial intelligence (AI)/machine learning (ML) scoring layer further ranks candidates using sequence features, AlphaFold2 structures, protein-protein interaction networks, and genotype-phenotype data from ∼20,000 clinical isolates. Five targets showed consistent cross-study prioritization: LpxC, MurA, FabI, DapD, and NGFG_RS03485. Of these, only LpxC has experimental inhibitor validation against multidrug-resistant gonococci; the remaining candidates are supported computationally but lack biochemical confirmation. Bridging this validation gap remains the critical bottleneck. However, advances in pan-genome analysis, structural prediction, and generative AI for compound design provide a more tractable route to novel therapeutics than previously available.