CaliciBoost: Performance-Driven Evaluation of Molecular Representations for Caco-2 Permeability Prediction

2025-06-09

AI Summary

Accurate prediction of oral absorption in early drug discovery remains challenging due to insufficient predictive performance of existing models. Method: This study systematically evaluates eight classes of molecular representations—including 2D/3D descriptors, ECFP/MACCS fingerprints, and GNN embeddings—for Caco-2 permeability prediction. Leveraging AutoML frameworks (TPOT and H2O), we propose CaliciBoost, a novel ensemble model optimized for small-sample ADMET modeling. Contribution/Results: We demonstrate for the first time that 3D descriptors significantly reduce mean absolute error (MAE) by 15.73%. CaliciBoost achieves state-of-the-art performance on both the TDC and OCHEM benchmark datasets, yielding the lowest MAE. Furthermore, we establish a reusable, principled framework for molecular representation selection tailored to low-data ADMET tasks. This work provides both methodological guidance and open-source tools for high-confidence oral absorption prediction.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Caco-2 permeability serves as a critical in vitro indicator for predicting the oral absorption of drug candidates during early-stage drug discovery. To enhance the accuracy and efficiency of computational predictions, we systematically investigated the impact of eight molecular feature representation types including 2D/3D descriptors, structural fingerprints, and deep learning-based embeddings combined with automated machine learning techniques to predict Caco-2 permeability. Using two datasets of differing scale and diversity (TDC benchmark and curated OCHEM data), we assessed model performance across representations and identified PaDEL, Mordred, and RDKit descriptors as particularly effective for Caco-2 prediction. Notably, the AutoML-based model CaliciBoost achieved the best MAE performance. Furthermore, for both PaDEL and Mordred representations, the incorporation of 3D descriptors resulted in a 15.73% reduction in MAE compared to using 2D features alone, as confirmed by feature importance analysis. These findings highlight the effectiveness of AutoML approaches in ADMET modeling and offer practical guidance for feature selection in data-limited prediction tasks.

Problem

• Evaluating molecular representations for Caco-2 permeability prediction
• Comparing 2D/3D descriptors and AutoML techniques for accuracy
• Identifying optimal feature sets to reduce prediction error

Innovation

• Combined 2D/3D descriptors with AutoML
• Used PaDEL, Mordred, RDKit for prediction
• Achieved 15.73% MAE reduction with 3D