Digital discovery最新文献

{"title":"Coupling causality and interpretable machine learning to reveal the reaction coordinate of C–N coupling with a supramolecular Cu-calix[8]arene catalyst","authors":"R. A. Talmazan, J. Gamper, I. Castillo, T. S. Hofer and M. Podewitz","doi":"10.1039/D5DD00216H","DOIUrl":"10.1039/D5DD00216H","url":null,"abstract":"<p >Supramolecular 3d transition-metal catalysts are large, flexible systems with intricate interactions, resulting in complex reaction coordinates. To capture their dynamic nature, we developed a broadly applicable, high-throughput workflow, that leverages quantum mechanics/molecular mechanics molecular dynamics (QM/MM MD) in explicit solvent, to investigate a Cu(<small>I</small>)-calix[8]arene-catalysed C–N coupling reaction. The system complexity and high amount of data generated from sampling the reaction requires automated analyses. To identify and quantify the reaction coordinate from noisy simulation trajectories, we applied interpretable machine learning techniques (Lasso, Random Forest, Logistic Regression) in a consensus model, alongside dimensionality reduction methods (PCA, LDA, tICA). By employing a Granger Causality model, we move beyond the traditional view of a reaction coordinate, by defining it instead as a sequence of molecular motions leading up to the reaction.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2954-2971"},"PeriodicalIF":6.2,"publicationDate":"2025-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12421827/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145042346","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Autonomous organic synthesis for redox flow batteries via flexible batch Bayesian optimization","authors":"Clara Tamura, Heather Job, Henry Chang, Wei Wang, Yangang Liang and Shijing Sun","doi":"10.1039/D5DD00017C","DOIUrl":"https://doi.org/10.1039/D5DD00017C","url":null,"abstract":"<p >Traditional trial-and-error methods for materials discovery are inefficient to meet the urgent demands posed by the rapid progression of climate change. This urgency has driven the increasing interest in integrating robotics and machine learning into materials research to accelerate experimental learning. However, idealized decision-making frameworks to achieve maximum sampling efficiency are not always compatible with high-throughput experimental workflows inside a laboratory. For multi-step chemical processes, differences in hardware capacities can complicate the digital framework by introducing constraints on the maximum number of samples in each step of the experiment, hence causing varying batch sizes in variable selection within the same batch. Therefore, designing flexible sampling algorithms is necessary to accommodate the multi-step synthesis with practical constraints unique to each high-throughput workflow. In this work, we designed and employed three strategies on a high-throughput robotic platform to optimize the sulfonation reaction of redox-active molecules used in flow batteries. Our strategies adapt to the multi-step experimental workflow, where their formulation and heating steps are separate, causing varying batch size requirements. By strategically sampling using clustering and mixed-variable batch Bayesian optimization, we were able to iteratively identify optimal conditions that maximize the yields. Our work presents a flexible approach that allows tailoring the machine learning decision-making to suit the practical constraints in individual high-throughput experimental platforms, followed by performing resource-efficient yield optimization using available open-source Python libraries.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2737-2751"},"PeriodicalIF":6.2,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00017c?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Adaptive subspace Bayesian optimization over molecular descriptor libraries for data-efficient chemical design","authors":"Farshud Sorourifar, Thomas Banker and Joel A. Paulson","doi":"10.1039/D5DD00188A","DOIUrl":"https://doi.org/10.1039/D5DD00188A","url":null,"abstract":"<p >The discovery of molecules with optimal functional properties is a central challenge across diverse fields such as energy storage, catalysis, and chemical sensing. However, molecular property optimization (MPO) remains difficult due to the combinatorial size of chemical space and the cost of acquiring property labels <em>via</em> simulations or wet-lab experiments. Bayesian optimization (BO) offers a principled framework for sample-efficient discovery in such settings, but its effectiveness depends critically on the quality of the molecular representation used to train the underlying probabilistic surrogate model. Existing approaches based on fingerprints, graphs, SMILES strings, or learned embeddings often struggle in low-data regimes due to high dimensionality or poorly structured latent spaces. Here, we introduce Molecular Descriptors with Actively Identified Subspaces (MolDAIS), a flexible molecular BO framework that adaptively identifies task-relevant subspaces within large descriptor libraries. Leveraging the sparse axis-aligned subspace (SAAS) prior introduced in recent BO literature, MolDAIS constructs parsimonious Gaussian process surrogate models that focus on task-relevant features as new data is acquired. In addition to validating this approach for descriptor-based MPO, we introduce two novel screening variants, which significantly reduce computational cost while preserving predictive accuracy and physical interpretability. We demonstrate that MolDAIS consistently outperforms state-of-the-art MPO methods across a suite of benchmark and real-world tasks, including single- and multi-objective optimization. Our results show that MolDAIS can identify near-optimal candidates from chemical libraries with over 100 000 molecules using fewer than 100 property evaluations, highlighting its promise as a practical tool for data-scarce molecular discovery.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2910-2926"},"PeriodicalIF":6.2,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00188a?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236690","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DropMicroFluidAgents (DMFAs): autonomous droplet microfluidic research framework through large language model agents","authors":"Dinh-Nguyen Nguyen, Raymond Kai-Yu Tong and Ngoc-Duy Dinh","doi":"10.1039/D5DD00306G","DOIUrl":"https://doi.org/10.1039/D5DD00306G","url":null,"abstract":"<p >Large language models (LLMs) have gained significant attention in recent years due to their impressive capabilities across various tasks, from natural language understanding to generation. Applying LLMs within specific domains requires substantial adaptation to account for the unique terminologies, nuances, and context-specific challenges inherent to those areas. Here, we introduce DropMicroFluidAgents (DMFAs) employing LLM agents to perform two key functions: (1) delivering focused guidance, answers, and suggestions specific to droplet microfluidics and (2) generating machine learning models to optimise and automate the design of droplet microfluidic devices, including the creation of code-based computer-aided design (CAD) scripts to enable rapid and precise design execution. To assess the accuracy of DMFAs in question–answering tasks, we compiled a dataset of questions with corresponding ground-truth answers and established an evaluation criterion. Experimental evaluations demonstrated that integrating DMFAs with the LLAMA3.1 model yielded the <em>highest accuracy of 76.15%</em>, underscoring the significant performance enhancement provided by agent integration. This effect was particularly pronounced when DMFAs were paired with the GEMMA2 model, resulting in <em>a 34.47% improvement in accuracy</em> compared to the standalone GEMMA2 configuration. For evaluating the performance of DMFAs in design automation, we utilized an existing dataset on flow-focusing droplet microfluidics. The resulting machine learning model demonstrated <em>a coefficient of determination of approximately 0.96</em>. To enhance usability, we developed a streamlined graphical user interface (GUI) that offers an intuitive and effective means for users to interact with the system. This study demonstrates the effective use of LLM agents in droplet microfluidics research as powerful tools for automating workflows, synthesising knowledge, optimising designs, and interacting with external systems, bringing a significant transformation to the field of digital discovery. DMFAs is capable of transforming them into closed-loop digital discovery platforms that encompass literature synthesis, hypothesis generation, autonomous design, execution in self-driving laboratories, analysis of results, and the generation of new hypotheses. These capabilities enable their application across education and industrial support, driving greater efficiency in scientific discovery and innovation.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2827-2851"},"PeriodicalIF":6.2,"publicationDate":"2025-08-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00306g?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236717","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating mechanistic modelling with Bayesian optimisation: accelerated self-driving laboratories for RAFT polymerisation","authors":"Clarissa. Y. P. Wilding, Richard. A. Bourne and Nicholas. J. Warren","doi":"10.1039/D5DD00258C","DOIUrl":"https://doi.org/10.1039/D5DD00258C","url":null,"abstract":"<p >Discovery of sustainable, high-performing materials on timescales to meet societal needs is only going to be achieved with the assistance of artificial intelligence and machine learning. Herein, a Bayesian optimisation algorithm is trained using <em>in silico</em> reactions facilitated by a new mechanistic model for reversible addition fragmentation chain transfer polymerisation (RAFT). This subsequently informs experimental multi-objective self-optimisation of RAFT polymerisation using an automated polymerisation platform capable of measuring the critical algorithm objectives (monomer conversion and molecular weight distribution) online. The platform autonomously identifies the Pareto-front representing the trade-off between monomer conversion and molar mass dispersity with a reduced number of reactions compared to the equivalent fully experimental optimisation process. This model-informed AI approach provides opportunities for much more sustainable and efficient discovery of polymeric materials and provides a benchmark for other complex chemical systems.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2797-2803"},"PeriodicalIF":6.2,"publicationDate":"2025-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00258c?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236715","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Extraction of chemical synthesis information using the World Avatar","authors":"Simon D. Rihm, Fabio Saluz, Aleksandar Kondinski, Jiaru Bai, Patrick W. V. Butler, Sebastian Mosbach, Jethro Akroyd and Markus Kraft","doi":"10.1039/D5DD00183H","DOIUrl":"https://doi.org/10.1039/D5DD00183H","url":null,"abstract":"<p >This work presents a generalisable process that transforms unstructured synthesis descriptions of metal–organic polyhedra (MOPs) – a class of organometallic nanocages – into machine-readable, structured representations, integrating them into The World Avatar (TWA), a universal knowledge representation encompassing physical, abstract, and conceptual entities. TWA makes use of knowledge graphs and semantic agents. While previous work established rational design principles for MOPs in the context of TWA, experimental verification remains a bottleneck due to the lack of accessible and structured synthesis data. However, synthesis information in the literature is often sparse, ambiguous, and embedded with implicit knowledge, making direct translation into structured formats a significant challenge. To achieve this, a synthesis ontology was developed to standardise the representation of chemical synthesis procedures by building on existing standardisation efforts. We then designed an LLM-based pipeline with advanced prompt engineering strategies to automate data extraction and created workflows for seamless integration into a knowledge representation within TWA. Using this approach, we extracted and uploaded nearly 300 synthesis procedures, automatically linking reactants, chemical building units, and MOPs to related entities across interconnected knowledge graphs. Over 90% of publications were processed successfully through the fully automated pipeline without manual intervention. The demonstrated use cases show that this framework supports chemists in designing and executing experiments and enables data-driven retrosynthetic analysis, laying the groundwork for autonomous, knowledge-guided discovery in reticular chemistry.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2893-2909"},"PeriodicalIF":6.2,"publicationDate":"2025-08-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00183h?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Accelerating optimization of halide perovskites: two blueprints for automation","authors":"Hilal Aybike Can, Daniel Anthony Jacobs, Nicolas Fürst, Christophe Ballif and Christian Michael Wolff","doi":"10.1039/D5DD00110B","DOIUrl":"10.1039/D5DD00110B","url":null,"abstract":"<p >The fine-tuning of halide perovskite materials for both performance and stability calls for innovative tools that streamline high-throughput experimentation. Here, we present two complementary systems designed to accelerate the development of solution-processed thin-film semiconductors. HITSTA (High-Throughput Stability Testing Apparatus) is a robust, cost-effective platform for optical characterization and accelerated aging, built around a repurposed 3D printer. It accommodates up to 49 thin-film samples, subjecting them to temperatures up to 110 °C and light intensities of 2.2 suns while continuously monitoring their absorptance and photoluminescence. ROSIE (Robotic Operating System for Ink Engineering) is a liquid-handling robot constructed from a hobbyist robotic arm and a syringe pump, enabling precise and automated ink formulation. We detail the design and operation of both systems, providing guidelines for their replication. To demonstrate their capabilities, we present a case study in which ROSIE and HITSTA are used to investigate the aging of mixed-cation, mixed-halide inorganic perovskites. Together, these systems form a powerful toolkit for accelerating the optimization of solution-processable thin-films <em>via</em> high-throughput experimentation.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2804-2815"},"PeriodicalIF":6.2,"publicationDate":"2025-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12381630/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144980892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Generative design of functional organic molecules for terahertz radiation detection","authors":"Zsuzsanna Koczor-Benda, Shayantan Chaudhuri, Joe Gilkes, Francesco Bartucca, Liming Li and Reinhard J. Maurer","doi":"10.1039/D5DD00106D","DOIUrl":"https://doi.org/10.1039/D5DD00106D","url":null,"abstract":"<p >Plasmonic nanocavities are molecule-nanoparticle junctions that offer a promising approach to upconvert terahertz radiation into visible or near-infrared light, enabling nanoscale detection at room temperature. However, the identification of molecules with strong terahertz-to-visible frequency upconversion efficiency is limited by the availability of suitable compounds in commercial databases. Here, we employ the generative autoregressive deep neural network, G-SchNet, to perform property-driven design of novel monothiolated molecules tailored for terahertz radiation detection. To design functional organic molecules, we iteratively bias G-SchNet to drive molecular generation towards highly active and synthesizable molecules based on machine learning-based property predictors, including molecular fingerprints and state-of-the-art neural networks. We study the reliability of these property predictors for generated molecules and analyze the chemical space and properties of generated molecules to identify trends in activity. Finally, we filter generated molecules and plan retrosynthetic routes from commercially available reactants to identify promising novel compounds and their most active vibrational modes in terahertz-to-visible upconversion.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2852-2863"},"PeriodicalIF":6.2,"publicationDate":"2025-08-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00106d?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Moment of inertia as a simple shape descriptor for diffusion-based shape-constrained molecular generation","authors":"Denis Sapegin, Fedor Bakharev, Dmitriy Krupenya, Azamat Gafurov, Konstantin Pildish and Joseph C. Bear","doi":"10.1039/D5DD00318K","DOIUrl":"https://doi.org/10.1039/D5DD00318K","url":null,"abstract":"<p >The article introduces <em>MLConformerGenerator</em>, a machine-learning framework for shape-constrained molecular generation that combines an Equivariant Diffusion Model (EDM), guided by a compact shape descriptor based on the principal components of the moment of inertia tensor, and a Graph Convolutional Network (GCN) model for bond prediction. The compact yet informative descriptor provides concise representation of molecular shape, enabling scalable learning from large datasets and synthetic conformers generated from 2D molecular inputs. The use of a GCN for bond prediction is evaluated in comparison to deterministic methods. The suggested approach provides an ability to fine-tune the model to generate datasets with chemical-feature distributions closely matching those of target datasets of real conformers. The proposed model supports generation conditioned on both explicit conformers and arbitrary shapes, offering flexibility for applications such as dataset augmentation and structure-based molecule design. Trained on over 1.6 million molecules, the model demonstrates the ability to generate chemically valid, structurally diverse molecules that conform to target shape constraints. It achieves an average shape similarity of 0.53 to a reference conformer, with peak similarity exceeding 0.9 – a performance comparable to that of analogous models relying on more complex descriptors. The results show that integrating physically grounded descriptors with modern generative architectures provides a robust and effective strategy for shape-constrained molecular design.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2927-2941"},"PeriodicalIF":6.2,"publicationDate":"2025-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00318k?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancing vanadium redox flow battery analysis: a deep learning approach for high-throughput 3D visualization and bubble quantification","authors":"André Colliard-Granero, Kangjun Duan, Roswitha Zeis, Michael H. Eikerling, Kourosh Malek and Mohammad J. Eslamibidgoli","doi":"10.1039/D5DD00158G","DOIUrl":"https://doi.org/10.1039/D5DD00158G","url":null,"abstract":"<p >This work harnesses deep learning to expedite analyses of research data for vanadium redox flow batteries. Recent studies have highlighted the significance of analyzing bubbles within vanadium redox flow batteries. The investigation of these bubbles had remained elusive in direct imaging until advancements in cell design facilitated their observation through synchrotron X-ray tomography. Yet, the considerable volume of slices per tomograph and the complexity of the features present challenges for analyzing bubbles. To tackle this issue, we propose a deep learning-based framework that allows experimentalists to conduct high-throughput analyses based on synchrotron X-ray tomographic images of vanadium redox flow batteries. We conducted a benchmarking study on various U-Net configurations using a dataset that includes three complete volumes. These volumes represent different cell configurations and encompass 2294 annotated images. Through a multi-class semantic segmentation approach, we aimed to identify four distinct classes, such as bubbles, electrolytes, membranes, and gaskets. The optimal model achieved a precision of 98%, a recall of 97%, and an F1-score of 97% on the validation set. Following segmentation, the framework facilitates rapid differentiation of electrodes, quantification of bubble volume, individual bubble shape analysis, generation of 2D bubble density maps, and calculation of membrane blockage. All results are readily accessible for interactive, on-site visualization within a 3D environment. The openly available software allows users to engage with the data in a comprehensive and intuitive manner. For access, please visit the following GitHub repository: https://github.com/andyco98/UTILE-Redox.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 10","pages":" 2724-2736"},"PeriodicalIF":6.2,"publicationDate":"2025-08-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2025/dd/d5dd00158g?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145236712","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}