AntiBac-Pred

AntiBac-Pred predicts antibacterial activity of chemical structures using (Q)SAR models trained on ChEMBL MIC data and supplemented with PASS (Prediction of Activity Spectra for Substances) predictions to support selection of compounds active below 10,000 nM against specified bacterial strains.

Key Features:

• Data-Driven Predictions: Uses ChEMBL bioactivity records, including minimum inhibitory concentrations (MICs) for compounds tested against 1,386 bacteria, to inform model training.
• (Q)SAR Modeling: Implements quantitative structure–activity relationship ((Q)SAR) approaches to predict antibacterial activity from chemical structures.
• Strain-Specific Classification: Classifies compounds as inhibitors or non-inhibitors for a panel of 353 bacterial strains, including resistant and non-resistant variants.
• Threshold-Based Prediction: Predicts potential inhibitory activity at concentrations below 10,000 nM as a practical activity threshold.
• Integration with PASS: Incorporates predictions from PASS (Prediction of Activity Spectra for Substances) to augment model-based assessments.

Scientific Applications:

• Rational Drug Design: Prioritizes chemical candidates for medicinal chemistry optimization based on predicted antibacterial activity and MIC thresholds.
• Screening Optimization: Narrows candidate sets for experimental testing by identifying compounds predicted to be active against specific bacterial strains.
• Mechanism Exploration: Supports identification of compounds with potential novel modes of antibacterial action through activity profiling across strains.

Methodology:

Quantitative structure–activity relationship ((Q)SAR) models trained on ChEMBL MIC data and augmented with PASS predictions are used to classify compounds as inhibitors or non-inhibitors for specified bacterial strains and to predict activity below 10,000 nM.