Concept formation in transformer language models is depth-extended, not a single-layer event: concepts emerge gradually across a contiguous region of the residual stream. Mechanistic interpretability methods identify the single layer of peak class separation -- the "best layer" -- capturing a snapshot rather than the process itself. We introduce the Concept Allocation Zone (CAZ): the depth interval within which a concept becomes measurably separable, the region allocated to its geometric expression. We formalize the CAZ through three layer-wise metrics (Separation, Concept Coherence, Concept Velocity) and derive principled boundary detection without manual layer sweeps. A CAZ is not a concept: it is the depth region within which the model organizes its geometry to make a concept separable. A single concept typically participates in multiple CAZes; multiple concepts may share one. Empirical validation across 34 models from 8 architectural families and 7 concepts reveals that the separation curve S(l) is frequently multimodal. A scored detector uncovers "gentle CAZes" -- subtle allocation regions invisible to standard peak detection but causally active in 93-100% of cases under ablat
Product concept evaluation is a critical stage that determines strategic resource allocation and project success in enterprises. However, traditional expert-led approaches face limitations such as subjective bias and high time and cost requirements. To support this process, this study proposes an automated approach utilizing a large language model (LLM)-based multi-agent system (MAS). Through a systematic analysis of previous research on product development and team collaboration, this study established two primary evaluation dimensions, namely technical feasibility and market feasibility. The proposed system consists of a team of eight virtual agents representing specialized domains such as R&D and marketing. These agents use retrieval-augmented generation (RAG) and real-time search tools to gather objective evidence and validate concepts through structured deliberations based on the established criteria. The agents were further fine-tuned using professional product review data to enhance their judgment accuracy. A case study involving professional display monitor concepts demonstrated that the system's evaluation rankings were consistent with those of senior industry experts.
We develop a workflow within the ONIOM framework and demonstrate it on the hybrid computing system consisting of the supercomputer Fugaku and the Quantinuum Reimei trapped-ion quantum computer. This hybrid platform extends the layered approach for biomolecular chemical reactions to accurately treat the active site, such as a protein, and the large and often weakly correlated molecular environment. Our result marks a significant milestone in enabling scalable and accurate simulation of complex biomolecular reactions
Prior research has revealed that certain abstract concepts are linearly represented as directions in the representation space of LLMs, predominantly centered around English. In this paper, we extend this investigation to a multilingual context, with a specific focus on human values-related concepts (i.e., value concepts) due to their significance for AI safety. Through our comprehensive exploration covering 7 types of human values, 16 languages and 3 LLM series with distinct multilinguality (e.g., monolingual, bilingual and multilingual), we first empirically confirm the presence of value concepts within LLMs in a multilingual format. Further analysis on the cross-lingual characteristics of these concepts reveals 3 traits arising from language resource disparities: cross-lingual inconsistency, distorted linguistic relationships, and unidirectional cross-lingual transfer between high- and low-resource languages, all in terms of value concepts. Moreover, we validate the feasibility of cross-lingual control over value alignment capabilities of LLMs, leveraging the dominant language as a source language. Ultimately, recognizing the significant impact of LLMs' multilinguality on our res
Understanding complex biomolecular mechanisms requires multi-step reasoning across molecular interactions, signaling cascades, and metabolic pathways. While large language models(LLMs) show promise in such tasks, their application to biomolecular problems is hindered by logical inconsistencies and the lack of grounding in domain knowledge. Existing approaches often exacerbate these issues: reasoning steps may deviate from biological facts or fail to capture long mechanistic dependencies. To address these challenges, we propose a Knowledge-Augmented Long-CoT Reasoning framework that integrates LLMs with knowledge graph-based multi-hop reasoning chains. The framework constructs mechanistic chains via guided multi-hop traversal and pruning on the knowledge graph; these chains are then incorporated into supervised fine-tuning to improve factual grounding and further refined with reinforcement learning to enhance reasoning reliability and consistency. Furthermore, to overcome the shortcomings of existing benchmarks, which are often restricted in scale and scope and lack annotations for deep reasoning chains, we introduce PrimeKGQA, a comprehensive benchmark for biomolecular question ans
Biomolecular interactions underpin almost all biological processes, and their rational design is central to programming new biological functions. Generative AI models have emerged as powerful tools for molecular design, yet most remain specialized for individual molecular types and lack fine-grained control over interaction details. Here we present ODesign, an all-atom generative world model for all-to-all biomolecular interaction design. ODesign allows scientists to specify epitopes on arbitrary targets and generate diverse classes of binding partners with fine-grained control. Across entity-, token-, and atom-level benchmarks in the protein modality, ODesign demonstrates superior controllability and performance to modality-specific baselines. Extending beyond proteins, it generalizes to nucleic acid and small-molecule design, enabling interaction types such as protein-binding RNA/DNA and RNA/DNA-binding ligands that were previously inaccessible. By unifying multimodal biomolecular interactions within a single generative framework, ODesign moves toward a general-purpose molecular world model capable of programmable design. ODesign is available at https://odesign.lglab.ac.cn ,
Generating large ensembles of candidate conformations is standard for improving biomolecular structure prediction. Yet aimless sampling is inefficient and costly, producing many redundant conformations with limited diversity, particularly for complex multimeric assemblies. Here, we present HelixFold-S1, a guided planning approach specifically designed to enhance the structural prediction of biomolecular complexes by strategically targeting the most informative regions of conformational space to produce accurate conformations. For each complex, predicted inter-chain contact probabilities serve as a blueprint of the conformational space, guiding computational effort toward higher-probability, low-redundancy contacts that constrain structure generation. Across diverse biomolecular complex benchmarks, HelixFold-S1 achieves markedly higher structural accuracy than traditional unguided methods while reducing sampling requirements by an order of magnitude. Predicted contact probabilities also provide a rough indicator of prediction difficulty and sampling utility. These results demonstrate that guided planning reshapes conformational exploration and enables more efficient and accurate str
Diffusion models excel at generating visually striking content from text but can inadvertently produce undesirable or harmful content when trained on unfiltered internet data. A practical solution is to selectively removing target concepts from the model, but this may impact the remaining concepts. Prior approaches have tried to balance this by introducing a loss term to preserve neutral content or a regularization term to minimize changes in the model parameters, yet resolving this trade-off remains challenging. In this work, we propose to identify and preserving concepts most affected by parameter changes, termed as \textit{adversarial concepts}. This approach ensures stable erasure with minimal impact on the other concepts. We demonstrate the effectiveness of our method using the Stable Diffusion model, showing that it outperforms state-of-the-art erasure methods in eliminating unwanted content while maintaining the integrity of other unrelated elements. Our code is available at https://github.com/tuananhbui89/Erasing-Adversarial-Preservation.
Structural assessment of biomolecular complexes is vital for translating molecular models into functional insights, shaping our understanding of biology and aiding drug discovery. However, current structure-based scoring functions often lack generalizability across diverse biomolecular systems. We present BioScore, a foundational scoring function that addresses key challenges -- data sparsity, cross-system representation, and task compatibility -- through a dual-scale geometric graph learning framework with tailored modules for structure assessment and affinity prediction. BioScore supports a wide range of tasks, including affinity prediction, conformation ranking, and structure-based virtual screening. Evaluated on 16 benchmarks spanning proteins, nucleic acids, small molecules, and carbohydrates, BioScore consistently outperforms or matches 70 traditional and deep learning methods. Our newly proposed PPI Benchmark further enables comprehensive evaluation of protein-protein complex scoring. BioScore demonstrates broad applicability: (1) pretraining on mixed-structure data boosts protein-protein affinity prediction by up to 40% and antigen-antibody binding correlation by over 90%;
Highly accurate biomolecular structure prediction is a key component of developing biomolecular foundation models, and one of the most critical aspects of building foundation models is identifying the recipes for scaling the model. In this work, we present SeedFold, a folding model that successfully scales up the model capacity. Our contributions are threefold: first, we identify an effective width-scaling strategy for the Pairformer to increase representation capacity; second, we introduce a novel linear triangular attention that reduces computational complexity to enable efficient scaling; finally, we construct a large-scale distillation dataset to substantially enlarge the training set. Experiments on FoldBench show that SeedFold outperforms AlphaFold3 on most protein-related tasks.
Biomolecular condensates composed of intrinsically disordered proteins (IDPs) are vital for proper cellular function, and their dysfunction is associated with diseases including neurodegeneration and cancer. Despite their biological importance, the precise physical mechanisms underlying condensate (dys)function are unclear, in part owing to the difficulties in understanding how biomolecular sequence patterns influence emergent condensate behaviours across relevant length and timescales. Here, through minimal physical modelling, we explain how IDP sequence patterning gives rise to nano-scale organisational heterogeneities in condensates. By applying our coarse-grained molecular-dynamics polymer model, which accounts for steric, attractive, and electrostatic interactions, we systematically quantify and map out the emergent morphological phases resulting from a wide range of sequence patterns. We demonstrate how sequences that enable local coil-to-globule transitions within regions of single polymers - driven by a competition between the preferred crowding densities of different regions in the sequence - also exhibit cluster formation in condensates. Overall, our work provides a conce
Large Language Models (LLMs), with their remarkable task-handling capabilities and innovative outputs, have catalyzed significant advancements across a spectrum of fields. However, their proficiency within specialized domains such as biomolecular studies remains limited. To address this challenge, we introduce Mol-Instructions, a comprehensive instruction dataset designed for the biomolecular domain. Mol-Instructions encompasses three key components: molecule-oriented instructions, protein-oriented instructions, and biomolecular text instructions. Each component aims to improve the understanding and prediction capabilities of LLMs concerning biomolecular features and behaviors. Through extensive instruction tuning experiments on LLMs, we demonstrate the effectiveness of Mol-Instructions in enhancing large models' performance in the intricate realm of biomolecular studies, thus fostering progress in the biomolecular research community. Mol-Instructions is publicly available for ongoing research and will undergo regular updates to enhance its applicability.
The convergence of statistical learning and molecular physics is transforming our approach to modeling biomolecular systems. Physics-informed machine learning (PIML) offers a systematic framework that integrates data-driven inference with physical constraints, resulting in models that are accurate, mechanistic, generalizable, and able to extrapolate beyond observed domains. This review surveys recent advances in physics-informed neural networks and operator learning, differentiable molecular simulation, and hybrid physics-ML potentials, with emphasis on long-timescale kinetics, rare events, and free-energy estimation. We frame these approaches as solutions to the "biomolecular closure problem", recovering unresolved interactions beyond classical force fields while preserving thermodynamic consistency and mechanistic interpretability. We examine theoretical foundations, tools and frameworks, computational trade-offs, and unresolved issues, including model expressiveness and stability. We outline prospective research avenues at the intersection of machine learning, statistical physics, and computational chemistry, contending that future advancements will depend on mechanistic inducti
Understanding and designing biomolecules, such as proteins and small molecules, is central to advancing drug discovery, synthetic biology and enzyme engineering. Recent breakthroughs in artificial intelligence have revolutionized biomolecular research, achieving remarkable accuracy in biomolecular prediction and design. However, a critical gap remains between artificial intelligence's computational capabilities and researchers' intuitive goals, particularly in using natural language to bridge complex tasks with human intentions. Large language models have shown potential to interpret human intentions, yet their application to biomolecular research remains nascent due to challenges including specialized knowledge requirements, multimodal data integration, and semantic alignment between natural language and biomolecules. To address these limitations, we present InstructBioMol, a large language model designed to bridge natural language and biomolecules through a comprehensive any-to-any alignment of natural language, molecules and proteins. This model can integrate multimodal biomolecules as the input, and enable researchers to articulate design goals in natural language, providing bi
The search for evidence of past life on Mars presents a tremendous challenge that requires the usage of very advanced robotic technologies to overcome it. Current digital microscopic imagers and spectrometers used for astrobiological examination suffer from limitations such as insufficient resolution, narrow detection range, and lack of portability. To overcome these challenges, this research study presents modifications to the Phoenix rover to expand its capability for detecting biosignatures on Mars. This paper examines the modifications implemented on the Phoenix rover to enhance its capability to detect a broader spectrum of biosignatures. One of the notable improvements comprises the integration of advanced digital microscopic imagers and spectrometers, enabling high-resolution examination of soil samples. Additionally, the mechanical components of the device have been reinforced to enhance maneuverability and optimize subsurface sampling capabilities. Empirical investigations have demonstrated that Phoenix has the capability to navigate diverse geological environments and procure samples for the purpose of biomolecular analysis. The biomolecular instrumentation and hybrid ana
Biomolecular condensates constitute a newly recognized form of spatial organization in living cells. Although many condensates are believed to form as a result of phase separation, the physicochemical properties that determine the phase behavior of heterogeneous biomolecular mixtures are only beginning to be explored. Theory and simulation provide invaluable tools for probing the relationship between molecular determinants, such as protein and RNA sequences, and the emergence of phase-separated condensates in such complex environments. This review covers recent advances in the prediction and computational design of biomolecular mixtures that phase-separate into many coexisting phases. First, we review efforts to understand the phase behavior of mixtures with hundreds or thousands of species using theoretical models and statistical approaches. We then describe progress in developing analytical theories and coarse-grained simulation models to predict multiphase condensates with the molecular detail required to make contact with biophysical experiments. We conclude by summarizing the challenges ahead for modeling the inhomogeneous spatial organization of biomolecular mixtures in livin
We consider approaches for controlling the buildup of noise by design of gates for chemical and biomolecular computing, in order to realize stable, scalable networks for multi-step information processing. Solvable rate-equation models are introduced and used to illustrate several recently developed concepts and methodologies. We also outline future challenges and possible research directions.
The period and amplitude of biomolecular oscillators are functionally important properties in multiple contexts. For a biomolecular oscillator, the overall constraints in how tuning of amplitude affects period, and vice versa, are generally unclear. Here we investigate this co-variation of the period and amplitude in mathematical models of biomolecular oscillators using both simulations and analytical approximations. We computed the amplitude-period co-variation of eleven benchmark biomolecular oscillators as their parameters were individually varied around a nominal value, classifying the various co-variation patterns such as a simultaneous increase/ decrease in period and amplitude. Next, we repeated the classification using a power norm-based amplitude metric, to account for the amplitudes of the many biomolecular species that may be part of the oscillations, finding largely similar trends. Finally, we calculate "scaling laws" of period-amplitude co-variation for a subset of these benchmark oscillators finding that as the approximated period increases, the upper bound of the amplitude increases, or reaches a constant value. Based on these results, we discuss the effect of differ
Decomposition of biomolecular reaction networks into pathways is a powerful approach to the analysis of metabolic and signalling networks. Current approaches based on analysis of the stoichiometric matrix reveal information about steady-state mass flows (reaction rates) through the network. In this work we show how pathway analysis of biomolecular networks can be extended using an energy-based approach to provide information about energy flows through the network. This energy-based approach is developed using the engineering-inspired bond graph methodology to represent biomolecular reaction networks. The approach is introduced using glycolysis as an exemplar; and is then applied to analyse the efficiency of free energy transduction in a biomolecular cycle model of a transporter protein (Sodium-Glucose Transport Protein 1, SGLT1). The overall aim of our work is to present a framework for modelling and analysis of biomolecular reactions and processes which considers energy flows and losses as well as mass transport.
Developing complex biomolecular workflows is not always straightforward. It requires tedious developments to enable the interoperability between the different biomolecular simulation and analysis tools. Moreover, the need to execute the pipelines on distributed systems increases the complexity of these developments. To address these issues, we propose a methodology to simplify the implementation of these workflows on HPC infrastructures. It combines a library, the BioExcel Building Blocks (BioBBs), that allows scientists to implement biomolecular pipelines as Python scripts, and the PyCOMPSs programming framework which allows to easily convert Python scripts into task-based parallel workflows executed in distributed computing systems such as HPC clusters, clouds, containerized platforms, etc. Using this methodology, we have implemented a set of computational molecular workflows and we have performed several experiments to validate its portability, scalability, reliability and malleability.