A Comparative Evaluation Framework Integrating Machine Learning and Deep Learning Models with ADME-Based Pharmacokinetic Assessment for HIV-Related Compounds
Aug 2026· Pharmaceuticals· Vol 19, pp. 1267· 0 citations· 58 references
Medicine
TL;DR
A comparative framework integrating classical ML, deep learning, graph-based models, and complementary ADME-based pharmacokinetic and drug-likeness assessment is presented to enable a comprehensive comparison of diverse molecular learning approaches.
Abstract
Background/Objectives: Predicting the bioactivity of HIV-related compounds is essential for early-stage drug discovery. However, most existing machine learning (ML) studies emphasize predictive performance while overlooking the predicted pharmacokinetic and drug-likeness properties of prioritized compounds. This study presents a comparative framework integrating classical ML, deep learning, graph-based models, and complementary ADME-based pharmacokinetic assessment. Methods: Twelve predictive models were evaluated using stratified five-fold cross-validation on the MoleculeNet HIV dataset under a unified experimental protocol. Model performance was assessed using multiple classification metrics together with statistical analysis. The highest-ranked compounds from the independent test set were further characterized using predicted ADME and drug-likeness properties. A representative compound (GDL1), prioritized by the GDL model, was subsequently evaluated by molecular docking against HIV-1 protease, HIV-1 integrase, and HIV-1 reverse transcriptase. Results: The graph-based GDL model achieved the highest ROC–AUC (0.956±0.015), followed by GRU (0.930±0.017) and RF (0.927±0.023). Statistical analysis indicated overall differences among model performances (Friedman test, p<0.001). However, Holm-corrected pairwise comparisons did not demonstrate statistically significant differences between the highest-performing models. Comparative ADME analysis showed that high predictive performance did not necessarily correspond to favorable predicted pharmacokinetic properties. Molecular docking suggested potential predicted binding interactions of the prioritized GDL1 compound with all three HIV-1 targets, with the most favorable predicted binding affinity observed for HIV-1 reverse transcriptase. Conclusions: The proposed framework enables a comprehensive comparison of diverse molecular learning approaches by integrating predictive performance with complementary predicted ADME, drug-likeness, and molecular docking analyses.
Accurate prediction of compound bioactivity is essential for accelerating antiviral drug discovery and reducing experimental costs. Machine learning (ML) methods have shown considerable promise in modeling structure–activity relationships and compound potency. In this study, we present an integrated ML framework for predicting IC50 and pIC50 values of compounds active against SARS-CoV-2, key indicators of antiviral potency. The proposed framework comprises three complementary approaches: (i) a regression model for quantitative IC50 prediction validated against experimental data; (ii) a classification model that categorizes compounds into active and inactive classes to support compound prioritization; and (iii) a multi-task neural network that jointly performs IC50 regression and activity classification, enhancing predictive performance and interpretability. A distinctive feature of this work is the incorporation of ligand efficiency (LE) as a criterion for activity classification, offering an alternative perspective on compound prioritization that has not been previously explored in SARS-CoV-2 bioactivity modeling. The proposed models demonstrate strong predictive capability, achieving a coefficient of determination (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$R^2$$\end{document}) of 0.77 using a neural network with feature selection, while the Random Forest classifier attains an accuracy, precision, and recall of approximately 0.92. These results highlight the potential of integrated regression, classification, and multi-task learning approaches as scalable and cost-effective tools for SARS-CoV-2 bioactivity prediction and antiviral drug discovery.
Aya I. Maiyza, Sohila Osama, Hanan A Hassan· BMC Bioinformatics· 0 citations
An integrated computational workflow combining explainable machine learning, virtual screening, molecular dynamics simulations, and binding free-energy calculations to identify novel inhibitors of this drug-resistant EGFR variant may support the development of new therapeutic strategies for overcoming resistance in EGFR-driven cancers.
Jurica Novak· International Journal of Mol...· 0 citations
Accurately predicting binding affinities between drugs and targets is crucial for drug discovery but remains challenging due to the complexity of modeling interactions between small drug and large targets. This research presents Dual modality feature fused-drug target affinity (DMFF-DTA), a model for drug-target affinity anticipation using dual-modality neural networks that considers both the sequence and graph structure of medicines and proteins. To facilitate more exact and efficient drug-target interaction modeling, the model incorporates a binding site-focused graph generation method for extracting binding information. Experimental results show that DMFF-DTA is far more effective than current state-of-the-art approaches. By outperforming state-of-the-art approaches by more than 8%, the model demonstrates remarkable generalizability to hitherto unexplored medicines and targets. The model's biological relevance is confirmed by the model interpretability analysis. This paper presents a reliable and understandable method for improving computational drug discovery by integrating multi-view protein and drug properties.
Ghazala Sultan, J. Vincent, Ratna Sahaya et al.· International Conference Com...· 0 citations
Quantitative prediction of inhibitor potency can accelerate early-stage drug discovery. Recently, data-driven approaches have gained widespread interest in drug discovery, as evidenced by a growing number of benchmarking challenges and open competitions. In this context, we developed a machine learning-based methodology that can find the most effective way of predicting IC50 values against ASK1 from SMILES, for "Jump AI(.py) 2025: 3rd AI Drug Discovery Competition", hosted by the Korea Pharmaceutical and Bio-Pharma Manufacturers Association (KPBMA) on the Dacon platform. Applying our methodology achieved the highest overall predictive performance among all participating teams. Beyond this competition setting, we present a compact SMILES-based modeling workflow comprising (i) a pre-trained encoder, (ii) regression models, (iii) data augmentation, and (iv) hyperparameter tuning. We systematically compared molecular representations from sequence- and graph-based models, including ChemBERTa-2 and MolCLR. Across encoder-regressor combinations, ChemBERTa-77 M-MLM embeddings paired with support vector regression (SVR) yielded the strongest predictive performance. Embedding-level mix-up augmentation and SVR hyperparameter tuning further improved predictive performance. Our findings highlight that careful SMILES preprocessing and encoder selection have a critical influence on IC50 values and provide a reproducible benchmark for single-target bioactivity prediction, thus contributing to a more efficient drug discovery process. Scientific Contribution In this study, we propose a machine learning methodology for predicting the IC50 values of ASK1 inhibitors from SMILES representations, with a systematic comparison of molecular encoders and regression models. Our results show that the use of suitable encoder-regressor pairs together with embedding-level mix-up augmentation improves model generalizability without requiring SMILES-level augmentation. This strategy would be particularly useful for settings with imbalanced labels or limited data, and could be applied more broadly to IC50 prediction for other kinase inhibitors.
Ju Hyung Lee, S. Choi, Utku Ozbulak et al.· Journal of Cheminformatics· 0 citations
In the early stages of drug discovery, predicting drug-target affinity is a crucial task. Due to the vast scale of genomic and chemical spaces, traditional biological methods are time-consuming, labor-intensive, and resource-demanding. As a result, machine learning-based computational methods have emerged to narrow down the pool of drug candidates. However, machine learning approaches still face several challenges in practical applications, particularly the scarcity of labeled samples and poor model generalization capability. To address these issues, this paper proposes a novel drug-target affinity prediction model, termed MetaBayes-DTA, based on an uncertainty-aware meta-learning framework. The model integrates the few-shot rapid adaptation capability of meta-learning with an uncertainty quantification mechanism to enhance prediction accuracy and reliability. MetaBayes-DTA is evaluated on two benchmark datasets, DAVIS and KIBA. Experimental results demonstrate that the proposed model outperforms existing methods.
Naihan Shi, Yanpeng Zhao, Wanying Li et al.· 2026 IEEE 27th China Confere...· 0 citations
This study systematically benchmark six ranking loss functions, including state-of-the-art listwise methods, and five types of molecular representations across two large-scale drug screening datasets, CTRP and PRISM, to demonstrate that listwise loss functions such as LambdaLoss and LambdaRank consistently excel in both early and overall ranking quality.
Faraz Sarmeili, Benyamin Ghahremani-Nezhad, Mohammad Khalilpour et al.· PLoS ONE· 0 citations