SciReasoner is introduced, a multimodal scientific foundation model for native structural reasoning across proteins, small molecules and inorganic crystals that connects accurate prediction with interpretable scientific inference.
Abstract
Structure-property relationships are foundational to biology, chemistry and materials science, where function, reactivity and physical response emerge from spatial, chemical and periodic organization. Mechanistically explaining these relationships requires interpreting structural evidence through scientific principles and physical constraints, from stereochemistry and bonding to symmetry, energetics and periodic order. However, applying artificial intelligence to this process presents a joint challenge of representation and reasoning: models must preserve domain-native structural information while showing how specific evidence supports predictions under these constraints. Here we introduce SciReasoner, a multimodal scientific foundation model for native structural reasoning across proteins, small molecules and inorganic crystals. SciReasoner discretizes coordinates, topologies and periodic connectivities into a unified structure-aware vocabulary, treating structural tokens as addressable evidence units during reasoning. In homology-controlled Gene Ontology prediction, SciReasoner improves Cellular Component annotation for low-homology and orphan-like proteins, increasing $F_{\max}$ from 0.42 to 0.55. In chemistry, it raises single-step retrosynthesis accuracy from 0.63 to 0.72 while generating fragment-level disconnection and precursor-verification traces. In materials science, its representations separate elemental and compound phases and resolve high- and low-band-gap regimes. Across 86 benchmarks, SciReasoner achieves state-of-the-art performance on 67 tasks. Double-blind expert evaluation rates its reasoning traces as preferred or at least comparable to those of a frontier large language model in 98% of cases. By making structure an inspectable substrate for reasoning under scientific constraints, SciReasoner connects accurate prediction with interpretable scientific inference.
RelAgent decomposes the task into three interpretable stages: entity extraction, substructure localization, and ontology-guided relationship reasoning, and then uses verifier agents to rank structurally plausible candidates to support fine-grained reasoning over molecular substructure.
Rubing Chen, Jiaxin Wu, C. Zhang et al.· Bioinformatics· 0 citations
Local chemical perception and property reasoning are both essential for understanding how molecular structure determines properties. Current LLM-based chemical reasoning methods either receive SMILES/molecular images together with descriptions of local motifs, or reason directly from molecular images. Neither approach enables the model to focus on chemically meaningful regions before reasoning. To address this gap, we propose Visual Latent Structural Reasoning (VLSR), an end-to-end framework that jointly learns localization and reasoning from molecular images. Central to our approach is a localize-then-reason strategy. VLSR first learns to locate chemically meaningful regions in a molecular image. It then reasons about their property effects in a compact latent workspace before producing the final answer. Under the same inference setup, this design achieves 9.6X higher throughput than a comparable textual-reasoning baseline.
A task-adaptive large reasoning model that integrates chemical knowledge through a synergistic multispecialist architecture, chain-of-thought supervision, and molecule-informed reinforcement learning is presented, demonstrating a versatile multitask framework for knowledge-guided molecular reasoning and design.
Pengfei Liu, Shuang Ge, Xiaobo Wang et al.· Journal of Physical Chemistr...· 0 citations
We model protein structural stability as a continuous scalar quantity defined over molecular geometry, referred to as the support field. Instead of treating stability as a discrete residue annotation or an empirical score, this representation characterizes protein folds through the combined effects of geometric organization, topological persistence, and local density. Based on this idea, we introduce Support Field Neural Representation Learning (SF-NRL), a topology-guided approach that integrates persistent homology(PH), spatial density estimation, and geometric deep learning to infer residue-wise support directly from protein structures. Persistent topological features are incorporated as structural constraints that modulate local support values across the fold, enabling a continuous description of structural reliability. Across diverse protein families, the inferred support field shows consistent agreement with independent indicators of structural stability and highlights low-support regions associated with conformational flexibility and weak structural integration. By embedding protein structures into a continuous stability landscape, SF-NRL provides an interpretable representation that complements structure prediction models and facilitates systematic identification of structural cores, flexible regions, and functionally relevant motifs. These results demonstrate that topology-informed field representations offer a generalizable and practically useful approach for analyzing protein stability and fold organization.
Accurate identification of interactions between protein residues and ligand functional groups is critical for understanding molecular recognition and guiding rational drug design. Existing deep learning approaches for protein–ligand interpretability typically rely on three-dimensional structural input or distance-based contact labels, which limit both their applicability and biological relevance. Here, we present LINKER, the first sequence-based model to predict residue-functional group interactions according to biologically defined interaction types, using only a protein sequence and the SMILES representation of the ligand. LINKER is trained via structure-supervised interaction learning, in which interaction labels are derived from three-dimensional protein–ligand complexes through functional group-based motif extraction. By representing ligands as ensembles of functional groups, the model emphasizes chemically meaningful substructures rather than mere spatial proximity. Importantly, LINKER requires only sequence-level input at inference, enabling large-scale applications in contexts where structural data are unavailable. Extensive experiments demonstrate that LINKER consistently outperforms established baselines, highlighting the utility of functional group abstractions and structure-based supervision for interpretable protein–ligand interaction prediction. Our source code is publicly available at: https://github.com/HySonLab/LINKER/.
Phuc Pham, Viet Thanh Duy Nguyen, Truong-Son Hy· Journal of Chemical Informat...· 1 citation
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