Digital discovery最新文献

{"title":"How digital is chemical research? insights from the second NFDI4Chem community survey on research data and FAIR workflows","authors":"Jochen Ortmeyer, Vitali Sidorin, Daniela Adele Hausen, Ann-Christin Andres, John Jolliffe, Theo Bender, Giacomo Lanza, Steffen Neumann, Oliver Koepler, Nicole Jung, Christoph Steinbeck, Johannes Liermann and Sonja Herres-Pawlis","doi":"10.1039/D5DD00584A","DOIUrl":"https://doi.org/10.1039/D5DD00584A","url":null,"abstract":"<p >Increasing digitalisation is revolutionising all scientific disciplines, but it poses a particular challenge in chemistry. Research data is generated, for example, during the synthesis of substances, the recording of spectra or the application of theoretical methods, and must therefore be appropriately documented, archived and made available for reuse. Traditionally, this happens in the form of paper-based laboratory notebooks. In order for the transition to the age of digital chemistry to succeed, a cultural change in the community is necessary. But how do chemists actually treat their research data? the recent NFDI4Chem survey provides answers on the progress of digitalisation. Moreover, we highlight how the consortium transported the community's needs into the second funding phase. This exemplifies a feedback-driven model for digital infrastructure design that could serve as a blueprint for other disciplines and national initiatives.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 2629-2638"},"PeriodicalIF":7.1,"publicationDate":"2026-05-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d5dd00584a?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148261443","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":"First annual Digital Discovery Emerging Investigators collection","authors":"None","doi":"10.1039/D6DD90016J","DOIUrl":"https://doi.org/10.1039/D6DD90016J","url":null,"abstract":"<p >A graphical abstract is available for this content</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 1962-1962"},"PeriodicalIF":6.2,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd90016j?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968349","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":"Contributors to the Digital Discovery Emerging Investigators collection 2025","authors":"None","doi":"10.1039/D6DD90017H","DOIUrl":"https://doi.org/10.1039/D6DD90017H","url":null,"abstract":"<p >A graphical abstract is available for this content</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 1963-1967"},"PeriodicalIF":6.2,"publicationDate":"2026-05-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd90017h?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968350","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":"Structured domain knowledge enables trustworthy materials science question-answering with large language models","authors":"Daegun Lee, Jiwoo Choi, Gyeong Hoon Yi, Seok Su Sohn, Byungju Lee and Donghun Kim","doi":"10.1039/D6DD00028B","DOIUrl":"https://doi.org/10.1039/D6DD00028B","url":null,"abstract":"<p >Large language models (LLMs) remain unreliable for materials science question answering because correct conclusions depend on detailed experimental conditions. Here, we show that a structured, domain-specific knowledge dataset is a critical prerequisite for trustworthy LLM-assisted question answering in materials science. Using water-splitting catalysis as a proof of concept, we curate the literature into a hierarchical, machine-queryable knowledge base encoding material synthesis, composition, and performance. This structured representation improves condition-aware retrieval and reduces context mismatches that commonly arise from superficial semantic similarity. Combined with query reformulation, it achieves 85.6% accuracy on 202 DOI-identification questions <em>versus</em> 21.3% for an unstructured baseline, while reducing operating cost by 39%. To assess broader free-form scientific question answering beyond exact-match retrieval, we further evaluate 202 descriptive questions using the RAGAS framework, which indicates more faithful, evidence-grounded answers. Together, these results show that structured domain knowledge can substantially improve the reliability of LLM-based materials science question answering.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2243-2253"},"PeriodicalIF":6.2,"publicationDate":"2026-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd00028b?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968330","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":"Identification of multi-transcriptomic prognostic biomarkers to explore natural therapeutics for lung cancer integrating machine learning","authors":"Md Ahad Ali, Hriddhi Sarker, Marguba Kamrun, Humaira Sheikh, Bilkis Akter Shifa, Siam Ahmed, Tarikul Islam, Sujoy Banik and Neeraj Kumar","doi":"10.1039/D6DD00045B","DOIUrl":"https://doi.org/10.1039/D6DD00045B","url":null,"abstract":"<p >Lung cancer remains the leading cause of cancer-related mortality worldwide, underscoring the urgent need for novel therapeutic strategies. Cyclin-dependent kinase 1 (CDK1), a central cell-cycle regulator, has emerged as an oncogenic driver and potential target in lung adenocarcinoma. This study aimed to integrate transcriptomics, machine learning (ML), and advanced <em>in silico</em> approaches to identify natural product-derived potential inhibitors targeting CDK1. To identify robust differentially expressed genes, first we analyzed four different datasets (GSE19804, GSE10072, GSE18842, and GSE10799). Protein–protein interaction network and topological analysis highlighted CDK1 as a primary key hub gene (pKHG) enriched in cell-cycle and p53 pathways. Target validation confirmed CDK1 overexpression, prognostic significance, immune infiltration links, and mutation associations. In addition, a collected library of 9667 natural phytocompounds was reduced through ML-based bioactivity (pIC50) prediction targeting pKHG to discover potential lead molecules. Then, the selected top lead molecules were considered for further evaluation <em>via</em> molecular docking, molecular dynamics simulations, ADMET analysis, and binding free-energy calculations (MM-GBSA). Among the selected phytochemicals, CID_14218027 (−7.69 kcal mol<small>⁻¹</small>), CID_487089 (−6.80 kcal mol<small>⁻¹</small>), and CID_174880 (−6.70 kcal mol<small>⁻¹</small>) showed the highest binding affinity score (GLIDE_XP score) and stable molecular interactions. Furthermore, MD simulations confirmed the conformational stability of ligand–protein complexes, supporting their potential as CDK1 inhibitors. This integrated omics-to-<em>in silico</em> pipeline identifies CDK1 as a robust therapeutic target and highlights natural product-derived inhibitors with favorable pharmacological and physicochemical properties. Therefore, these findings present a viable framework for accelerating precision drug discovery, with experimental validation underway. However, these findings are based solely on computational analyses and require further experimental validation to confirm CDK1 inhibitory activity, anticancer efficacy, and safety.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 6","pages":" 2520-2546"},"PeriodicalIF":7.1,"publicationDate":"2026-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd00045b?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148261440","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":"Discovery of hydrogen storage molecules using large language models and machine learning","authors":"Hassan Harb, Magali S. Ferrandon, Timothy A. Goetjen, Seryeong Lee, Omar K. Farha, Massimiliano Delferro and Rajeev Surendran Assary","doi":"10.1039/D6DD00102E","DOIUrl":"https://doi.org/10.1039/D6DD00102E","url":null,"abstract":"<p >Accelerating the discovery of new molecules with targeted properties is a central challenge in molecular design. In this contribution, we present an AI-driven molecular discovery framework that integrates Large Language Models (LLMs) for generative molecular design with Machine Learning (ML)-based screening to identify novel Liquid Organic Hydrogen Carrier (LOHC) candidates. Using the developed framework, LOHC molecules were systematically generated, evaluated, and refined iteratively, combining LLM-guided molecular generation and ML-predicted hydrogenation enthalpies (Δ<em>H</em>), under physicochemical property constraints such as optimal melting points (MP), desired hydrogen storage capacity (wt% H<small>₂</small>), and synthetic accessibility (SA) scores. This approach enabled the discovery of 42 new LOHC candidates in two distinct campaigns, one seeded with experimentally known and another with previously computationally identified LOHCs, respectively. Although we began with different numbers of starting molecules (31 <em>vs.</em> 7 seed molecules), both runs yielded a comparable number of viable candidates, suggesting an influence of chemically intuitive seed molecule selection for success. Selected LOHC molecules, such as 3-methyl pyridine, 1-ethylnapthalene, 1,1-diphenylethane, and benzofuran, were experimentally tested and compared with benchmark LOHCs (toluene and 9-ethylcarbazole) for hydrogenation using a series of commercial supported metal catalysts. The order of conversion into fully hydrogenated products at 200 °C was 3-methyl pyridine (100%) &gt; 9-ethyl carbazole (86.4%) &gt; 2,3-benzofuran (74%) &gt; 1,1-diphenylethane (66.9%) &gt; 1-ethylnapthalene (66.7%) &gt; toluene (57%), further validating the AI-guided molecular design. This study demonstrates promise of LLM-driven molecular design in conjunction with ML-based screening for accelerated discovery and design of molecules.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2089-2102"},"PeriodicalIF":6.2,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd00102e?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968359","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":"CReM-dock: de novo design of synthetically feasible structures guided by molecular docking","authors":"Guzel Minibaeva, Haolin Du, Finlay Clark, Julien Michel and Pavel Polishchuk","doi":"10.1039/D6DD00131A","DOIUrl":"https://doi.org/10.1039/D6DD00131A","url":null,"abstract":"<p >The <em>de novo</em> generation of chemical compounds represents a compelling strategy for the exploration of a significantly broader chemical space compared to traditional virtual screening methods. However, fragment-based approaches often encounter challenges related to the low synthetic accessibility of generated structures. In this study, we integrated the previously established fragment-based structure generator, CReM, with molecular docking techniques implemented in EasyDock to facilitate the exploration of chemical space. This novel methodology enables indirect control over the synthetic accessibility and diversity of the generated structures, enhances an objective function to yield compounds with desirable physicochemical properties, and maintains key protein–ligand interactions and ligand conformations. The structures generated through this approach exhibited a high degree of novelty and demonstrated competitiveness with those produced by leading methodologies in the field, as demonstrated by Absolute Binding Free Energy calculations. We illustrated the versatility of the developed approach through several case studies, showcasing its applicability to both <em>de novo</em> generation and fragment expansion tasks. The tool developed in this study is open-source and can be accessed at https://github.com/ci-lab-cz/crem-dock.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2271-2291"},"PeriodicalIF":6.2,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d6dd00131a?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968331","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":"Benchmarking physics-inspired machine learning models for transition metal complexes with diverse charge and spin states","authors":"Yuri Cho, Ksenia R. Briling, Yannick Calvino Alonso, Ruben Laplaza and Clemence Corminboeuf","doi":"10.1039/D5DD00571J","DOIUrl":"10.1039/D5DD00571J","url":null,"abstract":"<p >Physics-inspired machine learning (ML) models can be categorized into two classes: those relying solely on three-dimensional structure and those incorporating electronic information. In this work, we benchmark both classes for predicting quantum-chemical properties of transition metal complexes with diverse charge and spin states, using three complementary datasets. The evaluated methods include molecular representations (SLATM, FCHL, SOAP, and SPA<small>^H</small>M family) combined with kernel ridge regression, as well as geometric deep learning models (MACE and 3DMol). We examine how the inclusion of electronic information affects predictive accuracy across datasets and target properties. Models that incorporate electronic information consistently outperform purely structure-based models for properties whose distributions are strongly governed by electronic characters, such as spin-splitting energies and frontier orbital energies. In contrast, structure-only models perform well for predicting the HOMO–LUMO gap and dipole moment magnitude, whose distributions are relatively insensitive to electronic characteristics. Geometric deep learning models with charge and spin embeddings (MACE-QS and 3DMol-QS) show the highest overall accuracy, with 3DMol offering the best computational efficiency among the tested models. These results clarify when geometric information is sufficient and when incorporating electronic information becomes essential, providing practical guidance for selecting effective physics-based ML models for transition metal complexes.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2103-2119"},"PeriodicalIF":6.2,"publicationDate":"2026-04-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13150962/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147870717","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 sampling and SHAP interpretation of deposition-rates in bipolar HiPIMS","authors":"Alexander Wieczorek, Nathan Rodkey, Jan Sommerhäuser, Jason Hattrick-Simpers and Sebastian Siol","doi":"10.1039/D6DD00063K","DOIUrl":"10.1039/D6DD00063K","url":null,"abstract":"<p >High-power impulse magnetron sputtering (HiPIMS) offers considerable control over ion energy and flux, making it invaluable for tailoring the microstructure and properties of advanced functional coatings. However, compared to conventional sputtering techniques, HiPIMS suffers from reduced deposition rates. Many groups have begun to evaluate complex pulsing schemes to improve upon this, leveraging multi-pulse schemes (<em>e.g.</em> pre-ionization or bipolar pulses). Unfortunately, the increased complexity of these pulsing schemes has led to high-dimensionality parameter spaces that are prohibitive to classic design of experiments. In this work we evaluate bipolar HiPIMS pulses for improving deposition rates from Al and Ti sputter targets. Over 3000 process conditions were collected <em>via</em> autonomous Bayesian sampling over a 6-dimensional parameter space. The resulting machine-learning model was then interpreted using Shapley Additive Explanations (SHAP), to deconvolute complex process influences on deposition rates. This allows us to link observed variations in deposition rate to physical mechanisms such as back-attraction and plasma ignition. Insights gained from this approach were then used to target specific processes where the positive pulse components were expected to have the highest impact on deposition rates. However, in practice, only minimal improvements in deposition rate were achieved. In most cases, the positive pulse appears to be detrimental when placed immediately after the neg. pulse which we hypothesize relates to quenching of the afterglow plasma. The proposed workflow combining autonomous experimentation and interpretable machine learning is broadly applicable to the optimization of complex plasma processes, paving the way for physics-informed, data-driven advancements in coating technologies.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2221-2231"},"PeriodicalIF":6.2,"publicationDate":"2026-04-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13112526/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147790495","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":"Python-controlled, solvent-resistant fraction collector for automated flow synthesis","authors":"Hongchen Wang, Owen A. Meville, Harrison A. Mills, Monique Ngan, Jay R. Werber and Nipun Kumar Gupta","doi":"10.1039/D5DD00554J","DOIUrl":"https://doi.org/10.1039/D5DD00554J","url":null,"abstract":"<p >Flow chemistry has emerged as a powerful approach for high-throughput and automated synthesis workflows. However, downstream fraction collection remains a practical challenge: commercial systems are often expensive and difficult to integrate, while most low-cost or DIY alternatives lack compatibility with common organic solvents and real-time control. Here, we present a solvent-resistant, Python-controlled fraction collector built from low-cost, modular components (∼$1000). The system integrates programmable vial positioning and real-time volumetric monitoring to enable accurate, flow-independent fraction collection. Performance was validated across a range of flow rates (0.1–6 mL min<small>⁻¹</small>) and viscosities (0.45–500 cP), with consistent droplet mass and high reproducibility. It also achieves high separation fidelity, with sharp fluid transitions and negligible carryover between fractions under both aqueous and organic conditions. This open-source platform offers a robust, integrable solution for automated flow synthesis and supports broader adoption of continuous, data-rich experimentation in organic chemistry.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" 5","pages":" 2067-2073"},"PeriodicalIF":6.2,"publicationDate":"2026-04-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://pubs.rsc.org/en/content/articlepdf/2026/dd/d5dd00554j?page=search","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147968356","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}