By merging chemical heuristics with physical laws, DrugRPG effectively addresses the hallucination crisis, shifting generative SBDD from scoring-oriented optimization toward high-fidelity, realistic lead discovery.
Abstract
Generative artificial intelligence has shown great promise in structure-based drug design (SBDD), yet existing models often suffer from a fundamental crisis of "structural hallucinations", generating molecules with high binding scores that violate basic chemical principles or physical plausibility. Here we present DrugRPG, a physicochemically-grounded 3D molecule generation framework that bridges this gap by integrating deep-learned chemical priors with fundamental physical laws. DrugRPG introduces a cross-dimensional representation alignment objective, distilling knowledge from a chemical foundation model pre-trained on 600 million molecular similarities to ensure the generation of valid topologies and realistic pharmacophoric patterns. Simultaneously, a differentiable physics-guided sampling strategy, inspired by the Lennard-Jones potential, is applied during the reverse diffusion phase to dynamically mitigate steric clashes and enforce Van der Waals compatibility. Comprehensive benchmarking demonstrates that DrugRPG reduces severe steric clashes by 65.4% compared to the state-of-the-art baseline while maintaining competitive structural self-consistency. Crucially, DrugRPG achieves a 28.6% higher success rate in generating developable candidates that satisfy multi-objective criteria, including potency, stability, and synthetic feasibility. By merging chemical heuristics with physical laws, DrugRPG effectively addresses the hallucination crisis, shifting generative SBDD from scoring-oriented optimization toward high-fidelity, realistic lead discovery.
The results establish Molexar as a practical unified foundation for computational chemistry and drug-design workflows and establish Molexar as a practical unified foundation for computational chemistry and drug-design workflows.
Haoyu Lin, Yiyan Liao, Jinmei Pan et al.· 0 citations
MCTH (Monte Carlo Tree Hallucination), an inference-only framework that casts all-atom sequence-structure co-design as uncertainty-aware planning over hallucinated states from pretrained folding and inverse-folding models, with optional biophysical control within the same decision loop is introduced.
Xuefeng Liu, Mingxuan Cao, Xiao Luo et al.· 0 citations
PINS (Physics-Informed NMR Structure elucidation model), a generative framework that explicitly bridges the gap between spectral data and molecular topology by enforcing multiphysical priors, provides a trustworthy, automated strategy for decoding novel chemical structures in data-scarce regimes.
Current structure-based drug design generative models often struggle to faithfully recapitulate genuine ligand-protein binding interactions. Instead, under the coupling of implicit learning architectures and biased training data, they tend to learn spurious statistical correlations. To address this, we propose EIP-Diff (Explicit Interaction-Prompted Diffusion), an architecture featuring a novel explicit interaction-prompt embedding mechanism that is better suited for real-world target-specific drug design. This architecture replaces biased implicit learning with explicit, residue-level biological guidance, thereby promoting more fine-grained geometric fidelity and more precise interaction-aware conditioning. To fully realize the capabilities of EIP-Diff and provide a reliable basis for performance evaluation, we further constructed CrystalData set, which provides higher-fidelity and less-biased structural supervision than existing data sets. This explicit architecture markedly improves distribution consistency: even when trained on the crossdocked data set, EIP-Diff achieves the highest alignment with authentic pharmacological distributions among evaluated models. Training on CrystalData set further enhances this alignment and improves 3D geometric accuracy, while retaining strong controllability, high chemical space coverage, and near-perfect uniqueness. In addition, target-based validation on KAT6A and YTHDC1 confirmed that EIP-Diff accurately recapitulates native-like binding modes. Furthermore, in a real-world drug design task against IDO1, we successfully designed a novel lead compound with nanomolar potency (IC50 = 0.31 nM). These results demonstrate that the EIP-Diff architecture can explicitly leverage experimentally derived structural data and biologically meaningful interaction information for target-specific molecular generation, thereby enabling its effective application to real-world structure-based drug design.
Huabin Du, Mingyang Wang, M. Luo et al.· Journal of the American Chem...· 0 citations
This survey provides a comprehensive review of over 100 methods in 3D structure-based drug design (SBDD) and molecular optimisation, and proposes a unified taxonomy covering four primary generative paradigms: autoregressive models, diffusion models, flow matching, and Bayesian Flow Networks.
This review elucidates the paradigm shift in drug discovery from serendipitous exploration to rational, structure-driven polypharmacological molecular engineering, thereby providing a clear, structured guide for navigating the complexities of next-generation therapeutics.
Tianming Han, Zhijie Pan, Wenchi Ge et al.· 0 citations