Exploratory AI-Assisted In Silico Evidence That Bacterial Signaling Molecules May Occupy a Drug-Like Pharmacokinetic Space: A Preliminary Comparative Cheminformatics Analysis and Candidate Identification

The hypothesis that antibiotics evolved from or co-opted bacterial intercellular signaling molecules predicts that signaling-active compounds should exhibit physicochemical properties favorable for drug development. We explored this prediction through a comparative cheminformatics analysis of 74 bacterial signaling molecules — comprising 48 small-molecule diffusible signals, 17 peptide autoinducers and intracellular messengers, and 9 antibiotics with documented signaling activity — against 71 randomly selected drug-like molecules. Small-molecule signaling molecules showed significantly higher Lipinski Rule-of-5 compliance than random drugs (95.8% vs. 73.2%; Odds Ratio = 8.40, 95% CI: 1.86–38.1, p = 0.001; Bonferroni-adjusted p = 0.009), while peptide autoinducers and intracellular messengers showed near-zero compliance (5.9%), confirming that the enrichment is specific to small-molecule diffusible signals. Predicted membrane permeability was over 2-fold higher for small-molecule signals (72.9% vs. 32.4% classified as “Good”), and enrichment was observed across six of seven drug-likeness filters (Enrichment Factors 1.12–1.31). We identified 10 signaling-active molecules with favorable profiles whose direct antibiotic activity has not been systematically tested; six are free of PAINS alerts. This analysis was conducted with substantial AI assistance, molecular properties were compiled from databases rather than computed de novo, and results require independent verification with validated cheminformatics tools. We present these findings as a hypothesis-generating study to stimulate experimental testing of the signaling-first triage concept for antibiotic discovery. For all abbreviations and acronyms used throughout this manuscript, please refer to the Glossary section.