A unified deep generative framework for surrogate-guided molecular discovery across diverse molecular spaces.

From VAEs to Diffusion and LLMs: Modern Generative Models for Molecular Discovery

Generative models are emerging as a key technology for accelerating molecular discovery in drug design, materials science, and catalysis by enabling efficient exploration of the vast chemical space of possible molecules. Recent advances in deep generative modeling—including variational autoencoders (VAEs), diffusion models, flow matching methods, and autoregressive transformer-based approaches—have produced a diverse toolkit for generating molecular structures and optimizing their properties. However, these paradigms are often studied independently, leaving many machine learning researchers without a clear understanding of their connections, strengths, and limitations in molecular applications. This tutorial provides a unified introduction to modern generative modeling approaches for molecular generation, covering their theoretical foundations, algorithmic design, and practical considerations for molecular representations such as 1D SMILES strings, 2D molecular graphs, and 3D structures. While the tutorial primarily focuses on generative models for de novo molecular design, we also briefly discuss how similar modeling paradigms extend to reaction prediction and retrosynthesis. By presenting these models within a common framework, the tutorial aims to equip ML researchers and AI-for-science practitioners with a clear conceptual map of the generative modeling landscape for molecular discovery and identify emerging research opportunities in this rapidly evolving area.

MolGraphGAN: a graph transformer–adversarial framework for target-specific molecule generation

Dual-Attention Multimodal Framework for Molecular Property Prediction

A novel Dual-Attention Multimodal framework for Graphs and Sequence-based representations, so-called DAM-GS, which provides a promising solution for molecular property prediction with broad applications in drug discovery and computational molecular science.

HyperMolFusion: A Hypergraph-Enhanced Multi-Modal Fusion Framework for Accurate Molecular Property Prediction.

Molecular property prediction is a critical task in accelerating drug discovery. While deep learning has shown promise, prevailing single-modal methods struggle to integrate multi-source (e.g., atomic graph and molecular fingerprints), heterogeneous chemical knowledge, thereby failing to holistically represent molecular structures and capture the high-order synergistic interactions governing their functions. To address these challenges, we present HyperMolFusion, a hypergraph-enhanced multi-modal fusion model for molecular property prediction. Compared with traditional graphs limited to pairwise atomic bonds, HyperMolFusion models chemical motifs as hyperedges to explicitly capture high-order structural correlations and encode complex molecular interactions. The framework comprises three core representation learning modules: AtomConv for local atomic interaction learning via attention-enhanced message passing, HyperConv for motif-level high-order correlation extraction via hypergraph convolution with GRU gating, and a mixed molecular fingerprint module that adaptively integrates MACCS, PubChem, and Pharmacophore fingerprints. A chemically guided attention (CGA) mechanism then dynamically fuses these multi-level features into hierarchical molecular representations, alleviating over-smoothing and preserving structural information effectively. Evaluated on eight MoleculeNet benchmarks (covering regression and classification tasks), HyperMolFusion achieves promising performance. For regression, it achieves an RMSE of 0.611 in lipophilicity, 0.653 in ESOL, and 0.951 in FreeSolv. For classification, it achieves a ROC-AUC of 0.935 in ClinTox, 0.907 in BBBP, and 0.689 in SIDER. This work provides a systematic and effective solution for molecular property prediction: by holistically integrating atomic, motif, and global fingerprint information via hypergraph modeling, HyperMolFusion offers a more reliable computational tool to enhance the efficiency and accuracy of drug development pipelines.

Deep Learning Foundation Models for Low-Data Regimes from Classical Molecular Descriptors

This work proposes pretraining on low-noise, calculable molecular descriptors via supervised learning to obtain rich, highly transferable molecular representations and demonstrates this strategy with CheMeleon, a O(10M) parameter foundation model that enables directed message-passing neural networks to finally exceed the performance of classical methods in the low-data regime.

MARD-Mol: a hybrid autoregressive-diffusion paradigm for coarse-grained molecular modeling

Abstract Motivation Deep generative models have transformed drug molecule generation. However, molecules exhibit complex hierarchical structures, requiring models to simultaneously balance macroscopic topological coherence and microscopic chemical self-consistency. Although autoregressive (AR) and discrete diffusion paradigms are highly complementary, integrating their advantages within a unified architecture remains severely limited by traditional “atom-by-atom” fine-grained modeling. Results We propose MARD-Mol, a hybrid AR-diffusion framework based on motif-inspired units. By elevating the representation granularity from atoms to motif-inspired units and introducing a dual-stream hierarchical attention mechanism, it couples inter-unit AR global scaffold planning with intra-unit discrete diffusion generation. To support goal-directed drug discovery, we reformulate property optimization into an iterative “diagnose-and-repair” process, enabling targeted optimization of defective motifs while preserving the global scaffold. Extensive experiments demonstrate that MARD-Mol achieves an 86.0% Quality score in de novo generation and exhibits superior performance in fragment-constrained and multi-objective optimization, establishing a new paradigm for high-quality drug design. Availability and implementation The source code and datasets used in this study are available at GitHub: https://github.com/CSUBioGroup/MARD-Mol.