Proteome-Scale Mining and Multi-Objective Prioritization of Encrypted Antimicrobial Peptides with Experimental Validation

Assessing the translation of AI-prioritized genome-derived peptide fragments into validated antimicrobial candidates

The emergence of antibiotic-resistant pathogens such as Staphylococcus aureus demands accelerated antimicrobial discovery strategies. Artificial intelligence (AI) enables large-scale inference of candidate antimicrobial peptides (AMPs), yet experimental validation remains essential to determine whether predictions translate into biological function. Genome-guided mining, rather than unconstrained or randomly generated sequence exploration, offers a biologically grounded search space derived from organisms shaped by ecological and evolutionary pressures. Here, we evaluate this principle using Malassezia furfur, a skin-associated yeast that coexists with bacterial colonizers such as S. aureus, as a genomic source for AI-prioritized antimicrobial candidates. Candidate fragments were generated from two M. furfur genomes, filtered by physicochemical properties, prioritized with deep-learning AMP predictors, synthesized, and experimentally characterized. Selected peptides underwent cross-kingdom antimicrobial screening against S. aureus, combining kinetic growth and ultrastructural assays, complemented by in silico structural prediction, lipid-membrane interaction analysis, and human keratinocyte cytotoxicity evaluation. AI-guided genomic mining enriched biologically motivated sequence space for peptides with measurable antimicrobial activity, while revealing biases and generalizability limits of AI-based AMP inference. Closing the loop between genome-derived candidate generation, AI-based inference, synthesis, and functional characterization, this study provides an experimental assessment of model-guided AMP discovery and a reproducible route from computational prediction to validated antimicrobial candidates.

In silico identification and biophysical characterization of candidate antimicrobial peptides from the Indian marine microbiome targeting multidrug-resistant ESKAPE pathogens

The global health crisis of antimicrobial resistance necessitates the discovery of new antibacterial agents. Underexplored marine microbiomes, particularly from the biodiverse Indian coast, represent a rich potential source of antimicrobial peptides (AMPs). Targeting the urgent threat of multidrug-resistant ESKAPE pathogens, the present study aimed to computationally identify novel, membrane-active AMPs from these unique metagenomic datasets, with a focus on inhibiting Gram-negative bacteria. In this study, we computationally mined Indian marine high-resolution shotgun metagenomic datasets through quality filtering, de novo assembly, and small open reading frame prediction. An ensemble of six machine learning-based AMP prediction tools identified over 51,000 high-confidence candidate AMPs. Subsequent filtering based on physicochemical properties and AlphaFold3-predicted structures prioritized ten peptides with favourable membrane-active characteristics. Two lead candidates, c_AMP_1 and c_AMP_2, were subjected to all-atom molecular dynamics simulations within Gram-negative membrane mimetic models of Pseudomonas aeruginosa, Acinetobacter baumannii, and Klebsiella pneumoniae. Our simulations indicated distinct membrane interaction modes: c_AMP_1 adopted a stable, surface-associated α-helical orientation, while c_AMP_2 displayed a more flexible, membrane-inserting orientation in the simulations. Analysis of the MD simulations revealed distinct predicted peptide-membrane interaction profiles, characterized by specific hydrogen bonding patterns, peptide tilt angles, and membrane thinning, which collectively suggest differing biophysical interaction modes. Taken together, our work suggests the Indian marine microbiome as a promising reservoir for novel AMP candidates and suggests that an integrated computational pipeline – combining machine learning, structural biology, and biophysical simulation – may help prioritize candidate peptides for future experimental validation against critical pathogens.

Redundancy-aware AI-guided discovery and experimental validation of antimicrobial peptides for infected wound treatment.

Artificial intelligence (AI) is accelerating antimicrobial peptide (AMP) discovery, but prediction-centered workflows often overlook dataset redundancy, peptide synthesizability, and experimental anti-infective translation. Here, we developed a redundancy-aware AI-guided peptide discovery workflow integrating redundancy-controlled dataset construction, model interpretation, candidate screening, synthesis-linked experimental validation, and evaluation in an infected-wound model. A redundancy-retention (RR) dataset of 1861 peptides with E. coli MIC annotations was compared with CD-HIT-filtered CD60-CD90 datasets containing 439-1061 sequences. Redundancy control reshaped activity-density distributions, SHAP-derived feature dependence, and virtual-screening stringency. Screening 2.1 million random 13-mer peptides yielded 1763, 189, 157, 141, and 20 candidates from CD60, CD70, CD80, CD90, and RR workflows, respectively. Experimental synthesis and MIC testing showed that the RR-derived group had a higher mean crude yield and a higher hit rate against E. coli than the CD90-derived group (60% vs 20%; MIC ≤16 μM). The lead peptide A36 showed broad activity against the tested Gram-negative bacteria, inhibited drug-resistant clinical A. baumannii isolates, displayed low hemolysis and cytotoxicity, retained substantial integrity in serum and antibacterial activity after protease exposure, disrupted bacterial membranes, and reduced bacterial burden while promoting wound closure in an A. baumannii-infected wound model. These findings identify redundancy control as a practical factor influencing candidate selection, synthetic accessibility, and experimental hit recovery in AI-guided AMP discovery.

A Comprehensive In-Silico Pipeline for the Discovery of Non-toxic, Stable Antimicrobial Peptides From Databases for Targeting Multi-Drug Resistant K. pneumoniae

The rapid emergence of multidrug-resistant Klebsiella pneumoniae has significantly reduced the effectiveness of conventional antibiotics, highlighting the need for alternative therapeutic strategies. This study employed a comprehensive in silico pipeline to identify antimicrobial peptides (AMPs) targeting the essential DNA replication initiator protein DnaA. A total of 28,361 peptide sequences were collected from publicly available AMP databases and sequentially filtered based on peptide length, net charge, GRAVY score, instability index, antimicrobial activity, toxicity, hemolytic potential, aggregation propensity, sequence similarity and favorable amphipathic properties. Four peptides satisfied all selection criteria and were subjected to structural prediction, membrane-binding analysis, protein–DNA docking, protein–peptide docking, and Normal Mode Analysis. Protein–DNA docking identified the functional DNA-binding residues of DnaA, while peptide docking demonstrated that all four peptides interacted within this region. Peptide 3 exhibited the strongest predicted interaction, with a binding energy of −61.8±5.1, a buried surface area of 1151.7±30.8 Å2, and seven hydrogen bonds with key DnaA residues. Normal Mode Analysis further supported the structural stability of the peptide–protein complexes. These findings identify four promising AMP candidates targeting DnaA and provide a computational framework for peptide prioritization against multidrug-resistant K. pneumoniae. However, the proposed interactions remain computational predictions and require experimental validation.

Metagenomic screening of antimicrobial peptide candidates and isolation of two active peptides from bat gut bacteria

A Structure-Informed Graph Neural Network for Short Antimicrobial Peptide Discovery and Validation.

Short antimicrobial peptides (AMPs) are promising anti-infective agents due to their broad-spectrum antimicrobial activity, low likelihood of inducing resistance, and relative ease of synthesis and optimization. In this study, we developed GW-AMP, a residue-level graph neural network model that integrates sequence-derived descriptors with predicted structural information to facilitate the discovery of short AMPs. GW-AMP demonstrated robust and balanced classification performance in both cross-validation and an independent test set, outperforming several established AMP prediction models. Guided by model predictions, candidate peptides were selected and experimentally evaluated for antibacterial activity and hemolysis, confirming GW-AMP's effectiveness in identifying short AMPs with favorable activity and biocompatibility. Circular dichroism analysis further indicated that secondary-structure features and amphipathic distribution are closely linked to peptide potency and selectivity. Among the validated candidates, peptide 3 exhibited potent antibacterial activity, low hemolysis, and favorable in vivo efficacy and safety, supporting its potential as a lead compound for anti-infective development.