Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· pp. 12633-12642· 0 citations· 3 references
Abstract
Crystal Structure Prediction (CSP), the task of determining stable atomic arrangements from chemical composition alone, remains a central challenge in computational materials science with direct implications for accelerating materials discovery. While recent diffusion-based generative models achieve impressive results by conditioning on space-group symmetry information, this paradigm exposes three fundamental challenges: (1) unreliable symmetry inference from composition, (2) reliance on symmetry-only priors without comprehensive structural geometric guidance, and (3) prior-induced distribution shift caused by inaccurate or overly strong constraints. To address these challenges, we propose TGD-CSP, a three-stage generative framework that: (1) learns a cross-modal embedding space to retrieve structurally relevant templates directly from composition, thereby providing reliable symmetry priors; (2) guides diffusion-based generation via score-based conditioning that explicitly incorporates comprehensive geometric information from retrieved templates; and (3) fine-tunes the generative policy via reinforcement learning with an energy-based reward to alleviate prior-induced distribution shift and mitigate biased generation. % Evidence: Concrete results TGD-CSP achieves match rates of 74.56% and 75.25% on Perov-5 and MP-20, respectively, including a 42% relative improvement over state-of-the-art methods on Perov-5. It reduces RMSE to 0.0259 and 0.0258, respectively, while yielding structures with formation energies closely matching those of ground-truth structures. Our results demonstrate that TGD-CSP enables reliable generation from composition alone, significantly improving the practicality of generative models for real materials discovery.
Predicting the 3D structures of atomic systems is fundamental to advancing material science and drug discovery. While flow-matching models (, FlowDPO) have recently shown promise in this domain, their performance relies heavily on alignment with ground-truth coordinates via supervised preference learning. However, obtaining experimental labels for novel crystal phases or de novo proteins is prohibitively expensive, creating a bottleneck for structural modeling in data-scarce regimes. In this work, we propose (Atomic Policy Optimization), a fully unsupervised alignment framework that eliminates the need for ground-truth reference structures. APO adapts group-relative policy optimization to 3D atomic environments, utilizing a novel dual-reward mechanism: (i) a that reinforces the policy's dominant latent structural modes through eigen-decomposition of sample similarities, and (ii) a that enforces thermodynamic stability. Our framework enables the model to ``self-correct''by identifying physically plausible configurations within sampled groups. Extensive benchmarks on crystal and antibody structure prediction demonstrate that APO consistently outperforms fully supervised baselines, achieving a new state-of-the-art in match rates and structural fidelity. Furthermore, we show that APO effectively straightens probability paths, significantly improving inference efficiency. Our results suggest that intrinsic physical consistency can serve as a superior guide for alignment compared to noisy, supervised coordinate matching.
This review focuses on coordinate- and residue-frame-based diffusion approaches for generating protein structures, paying particular attention to geometric equivariance, conditioning strategies, all-atom modelling and interaction-aware design.
Wen-Ran Li, Xavier F. Cadet, David Medina-Ortiz et al.· International Journal of Mol...· 0 citations
Molecular crystal structure prediction (CSP) is important in pharmaceuticals, agrochemicals, and organic electronics, where subtle differences in molecular conformation and packing can strongly affect material properties. We present Packora, a flow-based generative model for molecular CSP that jointly predicts atomic coordinates and the lattice from molecular graphs. Packora supports multi-component and organometallic crystals and can condition on any subset of molecular conformers, stereochemical labels, and space-group information within a single model. Inspired by the CCDC CSP blind test, we evaluate generation and ranking separately, using generation to isolate generator quality and ranking to measure end-to-end performance under a common relaxation and ranking pipeline. We also systematically study architecture, training, conditioning, inference, and scaling, identifying an effective design based on cacheable pairwise reasoning, training objective and numerical solver choices, conditioning dropout, and balanced scaling of pairwise and single representations. Packora outperforms the baselines on both structure generation and ranking benchmarks, achieving the best matched-budget coverage across all six generation benchmarks, as well as higher experimental-form recovery, lower experimental-form ranks, and faster convergence in ranking.
Nayoung Kim, Kiyoung Seong, Sungsoo Ahn· 0 citations
Semantics-Preserving Exploratory Sampling (SPES), which integrates Graph Laplacian Spectral Noise to respect chain connectivity and Conditional Freedom Modulation to dynamically balance exploration with fidelity, enables diversity-enhanced generation without sacrificing geometric validity under the reported structural metrics, improving the empirical exploration--fidelity trade-off.
Zhihua Tian, Jiale Zhou, Rubo Wang et al.· Proceedings of the 32nd ACM...· 0 citations
Reliable structure-property modeling is crucial for accelerating materials discovery, where crystal graphs and structure-derived crystallographic descriptions provide complementary geometric and semantic information. Existing multimodal materials models primarily incorporate textual information through post-encoding fusion, latent-space alignment, or attention-based representation interaction mechanisms. However, in most cases, crystallographic semantics are introduced after structural encoding and therefore cannot directly guide the formation of atom-level crystal-graph representations. Here, we present Semantics-Augmented Geometric Encoder Network (SAGE-Net), a flexible multimodal framework that injects description-derived chemical and crystallographic semantics into geometric message passing. SAGE-Net introduces Semantic-Guided Message Passing (SGMP), which gates atom-level updates and enables crystallographic semantics to directly modulate local geometric interactions across multiple graph neural network (GNN) backbones. Across benchmarks covering bandgap, mechanical, transport-related properties, and synthesizability assessment, the SAGE-Net instantiated with different GNN backbones achieves the lowest MAE on eight out of ten JARVIS-DFT regression targets and delivers strong or highly competitive performance against both structure-based and multimodal baselines. For synthesizability assessment, the SAGE-Net demonstrate outstanding classification performance and high recall rates. Interpretability analysis unravels that SAGE-Net effectively captures physically interpretable crystallographic features, viz. space group, dimensionality, polyhedral environments, among others. Together, these results demonstrate SGMP-based SAGE-Net as a general and transferable framework for deeply integrated multimodal materials learning.
Guanghui Zhang, Yuxuan Yao, Kieran B. Spooner et al.· 0 citations
By directly resolving lattice periodicity and atomic columns, scanning transmission electron microscopy (STEM) offers rich structural cues, yet recovering a simulation-ready crystal structure from a single noisy STEM image remains an ill-posed inverse problem. In realistic acquisitions, corruption can obscure geometry-aligned cues, and composition-only structure generation is highly multimodal, yielding many plausible 2D slab candidates. We cast this task as a coupled inference problem that separates evidence recovery from structure inference under a minimal forward model. We introduce STEM2Crystal CoDiffusion (SCCD), a dual-diffusion framework that explicitly separates evidence recovery from structure inference and couples them via bidirectional feature exchange. SCCD comprises (i) an evidence diffusion branch that denoises a structure-aligned, mask-like evidence map conditioned on the noisy STEM image, and (ii) a crystal diffusion branch that jointly denoises the lattice and atomic coordinates under the given composition constraint. A bidirectional co-diffusion update enables iterative refinement: crystallographic context regularizes evidence denoising, while denoised image evidence provides noise-robust geometric cues that sharpen structure inference. For controlled evaluation, we release a large-scale synthetic benchmark with explicit composition constraints, controlled noise regimes, and projection-derived supervision signals. Across multiple complementary metrics and all noise levels, SCCD consistently outperforms strong baselines.
Guangyao Chen, Fengqi You· Proceedings of the 32nd ACM...· 0 citations