6 papers indexed here
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
Drug-induced liver injury (DILI) is a major cause of drug development failure and post-marketing withdrawal. Accurate computational prediction of hepatotoxicity is hindered by complex biological mechanisms and scarce labeled toxicity data. Although pretrained molecular language models like ChemBERTa perform well in molecular property prediction, their generalization ability for small-sample DILI prediction remains underexplored. Here, we systematically compared traditional molecular fingerprint-based machine learning methods and ChemBERTa-based models for DILI classification on the DILIst dataset. Canonical SMILES from PubChem were used to generate Morgan fingerprints and ChemBERTa embeddings. We evaluated Random Forest, XGBoost, full fine-tuning, frozen encoder transfer learning, and embedding-based classifiers under both random and scaffold data splits. Results showed that Morgan fingerprints combined with Random Forest achieved the best performance, with ROC-AUC of 0.783 and PR-AUC of 0.849 under random split. Scaffold split markedly degraded the performance of all models, indicating poor generalization to unseen chemical scaffolds. ChemBERTa embedding-based classifiers outperformed end-to-end fine-tuning, suggesting that pretrained representations are better used as fixed feature extractors under limited labeled DILI data. Further SHAP analysis detected key toxicity-related molecular fragments, and t-SNE showed insufficient latent-space separation between DILI-positive and negative compounds. Our results confirm that traditional fingerprint-based machine learning remains highly competitive for small-sample hepatotoxicity prediction, and this work provides a reliable computational framework for early drug safety assessment.
Molecular generation is a core task in drug discovery. Although existing deep generative models can produce valid molecular structures, they lack precise control over molecular properties such as lipophilicity (logP) and drug-likeness (QED). This paper proposes Property-Regularized Graph Variational Autoencoder (PR-GVAE), a conditional graph VAE that introduces Feature-wise Linear Modulation (FiLM) conditioning in the encoder and a dual-branch fusion architecture in the decoder for precise property control. Experiments on a filtered subset of ZINC-20 (approximately 52,000 molecules, at most 20 heavy atoms) across five models spanning graph-based, sequence-based, and fragment-assembly paradigms demonstrate that PRGVAE achieves a Condition Satisfaction Rate (CSR) for both properties of 78.3%, outperforming the unconditional Vanilla GraphVAE by 2.0× (39.4%) and the sequence-based SMILES C-VAE by 40 percentage points (38.3%). Ablation experiments reveal that decoder conditioning is the core driver of property control (removing it reduces CSR-Both by 26.6%), while encoder FiLM conditioning and dual-branch fusion exhibit additive contributions (simultaneous removal reduces CSR-Both by 30.8%). Visualization confirms that PR-GVAE generates structurally diverse drug-like molecules, providing an effective solution for on-demand molecular design.
Drug–target interaction (DTI) prediction is central to drug discovery, target identification, and drug repurposing. With the rapid growth of biomedical data and advances in artificial intelligence (AI), DTI prediction has shifted from docking, similarity-based inference, and hand-crafted features toward data-driven representation learning and interaction modeling. This review examines AI-driven DTI prediction methodologies, covering binding theories, task formulations, data representation, model design, translational applications, and unresolved challenges. First, we introduce classical molecular binding theories, including the lock-and-key model, induced fit, and conformational selection, highlighting the transition from static matching to dynamic interaction. Second, we summarize major DTI task settings, including binary interaction classification, binding affinity regression, and multitask prediction with uncertainty assessment. We then discuss data resources and multimodal representation approaches for drugs, target proteins, interaction labels, and auxiliary biomedical data, including molecular sequences, graph structures, 3D conformations, physicochemical properties, biological perturbation profiles, protein sequences and structures, and biomedical knowledge networks. Representative approaches are compared across orthogonal methodological dimensions, including input representation, encoder architecture, interaction-modeling mechanism, representation learning and pretraining, learning objective, prediction output, data acquisition or optimization strategy, and generalization setting. Finally, we outline DTI applications in disease target mining, compound virtual screening, affinity and selectivity optimization, complex structure prediction, and industrial drug discovery pipelines and discuss key challenges such as data distribution shifts, dynamic protein conformational variability, and insufficient experimental validation. This survey aims to provide a systematic reference for future algorithm design, mechanism exploration, and real-world drug discovery applications.
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.