Identification of chemical features for improved outer membrane permeation in mycobacteria using machine learning

2026-06-30 · Source: Machine learning : nature.com subject feeds · Field: Science & Research — Life Sciences & Biology, Health & Medical Research, Artificial Intelligence & Machine Learning · Depth: Expert, extended

Summary

A recent study identified chemical features for improved outer membrane permeation in mycobacteria, a critical factor for antibiotic efficacy against Mycobacterium tuberculosis (Mtb). Researchers used the bioorthogonal click chemistry-based PAC-MAN assay to profile 1,572 azide-tagged compounds in M. tuberculosis and M. smegmatis. Cheminformatics and a deep learning model, MycoPermeNet, identified that nitrogen-containing aromatic scaffolds, such as indole, imidazole, or pyrazole, positively correlate with mycomembrane permeation. MycoPermeNet achieved an R^2 of 0.51 on test data, accurately predicting permeability. Experimental validation across three molecule series (JSF-2985 analogues, W peptides, and Octyl tridecaptin A1 derivatives) confirmed these chemical features enhance permeation and, in some cases, anti-Mtb activity. This work provides a rational framework for designing more effective antitubercular drugs.

Key takeaway

For medicinal chemists developing antitubercular agents, prioritize chemical modifications that enhance mycomembrane permeation. Incorporating nitrogen-containing aromatic scaffolds, particularly indole, can significantly improve drug entry into Mycobacterium tuberculosis cells. Utilize predictive models like MycoPermeNet to guide lead compound derivatization and screen libraries, focusing on compounds with high predicted permeability to overcome this critical barrier and boost whole-cell activity.

Key insights

Indole and other nitrogen-containing aromatic scaffolds significantly enhance mycobacterial outer membrane permeation and anti-TB activity.

Principles

• Mycobacterial outer membrane permeation is scaffold-dependent.
• Aromatic nitrogen heterocycles, especially indole, promote permeation.
• ML models can predict mycomembrane permeability from chemical structure.

Method

The PAC-MAN assay, combined with cheminformatics and a deep learning model (MycoPermeNet), profiles and predicts mycomembrane permeation. A surrogate XGBoost model interprets physicochemical property influence.

In practice

• Incorporate indole or similar nitrogen-containing aromatic scaffolds into drug candidates.
• Use MycoPermeNet to guide lead compound derivatization for improved permeation.
• Prioritize drug candidates with predicted high mycomembrane permeability.

Topics

• Mycobacterium tuberculosis
• Antibiotic Permeation
• Machine Learning
• Cheminformatics
• Drug Discovery
• Indole Scaffolds

Code references

• Nevbarunegbe/Mycomembrane-permeability-project
• Nevbarunegbe/Mycomembrane-permeability-project

Best for: Research Scientist, AI Scientist, Machine Learning Engineer

Related on AIssential

• Convolutional neural networks quantify antibiotic resistance in Mycobacterium tuberculosis with diagnostic grade accuracy and predict treatment response — CNNs can predict antibiotic MICs from genomic data with clinical accuracy, leveraging diverse biological features.
• Probe Before You Edit: Probing-Guided Molecular Optimization for LLM Agents in Structure-Based Drug Design — Probing molecular edit responses guides LLM agents to jointly optimize binding affinity and druggability in drug design.
• DeepDrugDiscovery identifies blood–brain barrier permeable autophagy enhancers for Alzheimer’s disease — AI-driven screening can identify brain-permeable, mTOR-independent autophagy enhancers for Alzheimer's disease.
• Navigating molecular OOD-ness — A new metric quantifies chemical distribution shift to improve molecular machine learning generalization in drug discovery.
• BioMiner: A Multi-modal System for Automated Mining of Protein-Ligand Bioactivity Data from Literature — BioMiner automates protein-ligand bioactivity data extraction by separating semantic interpretation from chemical structure construction.
• APCyc: Property-Informed Design of Cyclic Peptides via Automated Cyclization — APCyc automates cyclic peptide design by inferring cyclization topology and optimizing multiple drug-like properties via Bayesian guidance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine learning : nature.com subject feeds.