Digital discovery最新文献

{"title":"RXNMapper-v2: updated validation and evaluation on expanded chemical space.","authors":"Yves Grandjean, Sacha Raffaud, Annie M Westerlund, Philippe Schwaller, Samuel Genheden, Jean-Louis Reymond","doi":"10.1039/d6dd00425c","DOIUrl":"https://doi.org/10.1039/d6dd00425c","url":null,"abstract":"<p><p>Atom-to-atom mapping (AAM) assigns reactant atoms to product atoms and is fundamental for reaction classification, template extraction, and synthesis planning. RXNMapper derives AAM from the attention weights of a transformer model trained by masked language modeling on unmapped reaction SMILES. However, its original model selection relied on USPTO-50K, a validation set later shown to contain erroneous mappings. Here we revisit RXNMapper using a curated validation set of 4'533 reactions from the dataset Pistachio. We retrained ALBERT-based RXNMapper models on USPTO-full and Pistachio and selected the best head, layer, and checkpoint on this new validation set. To improve the comparison of AAM algorithms, we developed an evaluation framework that removes ambiguous reference reactions and compares predictions based on reaction graph isomorphism rather than direct string matching, avoiding false negatives from inconsistent formats and symmetry-equivalent mappings. Across the golden dataset, Pistachio, and an internal AstraZeneca medicinal chemistry benchmark, the updated models outperform the original RXNMapper and LocalMapper, while rearrangements remain persistent challenges.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13678721/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149229439","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":"Commit: BitBIRCH: efficient clustering of large molecular libraries.","authors":"Ignacio Pickering, Krisztina Zsigmond, Kenneth López Pérez, Miroslav Lžičař, Ramón Alain Miranda-Quintana","doi":"10.1039/d5dd00493d","DOIUrl":"10.1039/d5dd00493d","url":null,"abstract":"<p><p>We present BitBIRCH-Lean, a fast, memory-efficient implementation of the BitBIRCH algorithm, designed for high-throughput clustering of huge molecular libraries (up to billions of drug-like molecules) on modern scientific workstations. BitBIRCH-Lean considerably improves on the original BitBIRCH implementation by incorporating dynamic types and bit-packed fingerprints inside the clustering tree. Most operations in BitBIRCH-Lean are efficiently performed on compressed data, and optional C++ extension accelerates the bottleneck calculations, providing up to 2× speedup. Benchmark tests against GPU-accelerated methods highlight BitBIRCH-Lean as an efficient alternative for processing vast amounts of molecules. We further demonstrate the versatility of this new package by showcasing a parallel, multi-round variant of the BitBIRCH algorithm that exploits the gains in efficiency to cluster hundreds of millions of molecules in minutes, while preserving the same cluster distribution and properties. The code is freely available at: https://github.com/mqcomplab/bblean.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13673950/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149217399","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":"Pharmacological stratification of public bioactivity databases: a reusable, OECD-anchored curation and benchmarking framework demonstrated for opioid receptors.","authors":"Manal A Nael, Laxman M Alakonda, Abhik Ghosh, Sara Jane Ward, Lee-Yuan Liu Chen, Anjali M Rajadhyaksha, Magid Abou-Gharbia, Khaled M Elokely","doi":"10.1039/d6dd00392c","DOIUrl":"https://doi.org/10.1039/d6dd00392c","url":null,"abstract":"<p><p>Public bioactivity databases are heterogeneous not only in measurement type, where binding affinities and functional potencies are reported on different scales, but in pharmacology: the same compound and target can carry agonist, antagonist, or inhibitor records measured through binding displacement, cAMP, β-arrestin, or [³⁵S]GTPγS readouts that quantify different biological events. Pooling these records produces models whose output is detached from any coherent pharmacological claim. Prior work has standardized bioactivity at scale and quantified the noise from mixing measurement types, but pharmacological mechanism and assay-readout class have not been treated as a primary axis of large-scale curation. This study presents an auditable, OECD-anchored framework that stratifies public records by action type and assay readout before modeling, converting heterogeneous data into externally validated, interpretable QSAR tasks that compose with existing standardization resources rather than replacing them. The framework is demonstrated on four opioid receptors (MOR, DOR, KOR, and nociceptin/orphanin FQ, NOP). Four public sources were reconciled into 72 148 merged records and 50 977 curated measurements spanning 19 585 compounds, each carrying auditable attributes for source agreement, endpoint meaning, pharmacology class, assay readout, and trust tier. Receptor-level binding tasks formed a compact benchmark with strong locked external performance, including KOR p<i>K</i> _i (<i>R</i> ² = 0.79, <i>n</i> = 798) and DOR p<i>K</i> _i (<i>R</i> ² = 0.77, <i>n</i> = 736). Pharmacology- and readout-resolved functional endpoints yielded externally validated strata that pooled labels would obscure, including an MOR antagonist functional-inhibition endpoint (<i>R</i> ² = 0.86, <i>n</i> = 110) and agonist potency endpoints for DOR, KOR, and MOR (<i>R</i> ² up to 0.81). Comparison against a fully pooled baseline shows that pooled models either match stratified models on coherent endpoints or reach a deceptively high <i>R</i> ² on functional-IC₅₀ endpoints by training predominantly on binding-displacement records, so the pooled number predicts affinity rather than functional activity. SHAP attribution indicates that binding and functional potency encode partially distinct structure-activity signals. The dataset contract, not model performance alone, defines the validity and scope of a QSAR claim, and stratification is a precondition for a functional model to support a defensible claim. Curation logic, derived tables, frozen data, and reproducibility artifacts are released.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-10-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13656078/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149186656","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 machine learning and molecular dynamics methods for the prediction of the thermophysical properties of organic liquids.","authors":"Egheosa Ogbomo, Fionn Carman, Daniele Dini, James P Ewen","doi":"10.1039/d6dd00260a","DOIUrl":"10.1039/d6dd00260a","url":null,"abstract":"<p><p>Accurate prediction of thermophysical properties is crucial for the rational design of lubricants and cooling fluids. We compile a database of approximately 4700 critically evaluated reference values for the density, viscosity, and thermal conductivity of 835 unique organic molecules in the C₁₀-C₅₀ range. We use this database to train and benchmark a range of machine learning (ML) architectures and molecular representations. XGBoost models trained on molecular descriptors are consistently among the most accurate across all three target properties. Out-of-distribution tests show that descriptor-space distance and the position of the predicted value within the training range identify higher-risk regions at the population level, rather than ordering the errors of individual molecules. Equilibrium molecular dynamics (EMD) simulations of eight representative molecules reproduce density accurately but show systematic, chemistry-dependent force-field errors for the viscosity and thermal conductivity. At an inference cost orders of magnitude lower than EMD simulations, the ML models are well suited to high-throughput screening of new candidate molecules for lubricants and other formulated products.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13647254/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149139227","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":"Systematic trends in water properties across Jacob's ladder density functionals.","authors":"Nam H Đào, Daniel S King, Frederick Stein, Xiaoyu Wang, Dongjin Kim, Gerit Brandenburg, Jürg Hutter, Bingqing Cheng","doi":"10.1039/d6dd00571c","DOIUrl":"10.1039/d6dd00571c","url":null,"abstract":"<p><p>Kohn-Sham density functional theory is the workhorse of computational materials science and chemistry, but the optimal choice of exchange-correlation approximation for a given system is often unclear. Traditional benchmark studies typically compare selected density functional approximations (DFAs) against high-level quantum chemistry results for small structures and static energies, rather than against experimental thermodynamic or response properties. Here, we show how machine learning interatomic potentials (MLIPs) allow for molecular dynamics simulations of these properties at scale, thereby providing a thermodynamic assessment for DFAs. Focusing on the well-studied but challenging case of bulk water at ambient conditions, we evaluate 11 DFAs spanning the top four rungs of Jacob's ladder (GGA, <i>meta</i>-GGA, hybrid, and double hybrid) with and without empirical dispersion corrections, using consistent basis set selections. Using MLIPs trained on each DFA, we compute infrared spectra, heat capacities, radial distribution functions, and diffusivities from classical molecular dynamics simulations. We find a general improvement up Jacob's ladder towards double hybrids, though there is substantial variation between DFAs within the same rung and across different properties. Dispersion corrections yield only a modest influence on the computed properties. We elucidate the similarities and hierarchy among DFAs, both based on correlating the DFA energies and forces for the same training configurations, and based on analyzing the finite-temperature observables. Our work thus demonstrates the power of MLIPs for finite-temperature benchmarking of electronic structure methods.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-09-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13636407/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149119337","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":"An accessible property classification framework to predict the solubility of functionalised naphthalenes and rylenes in organic solvents.","authors":"Connor R M MacDonald, Alex S Loch, Neil G Berry, Emily R Draper","doi":"10.1039/d6dd00275g","DOIUrl":"10.1039/d6dd00275g","url":null,"abstract":"<p><p>The development of functional materials such as multilayer films from supramolecular self-assembling materials is hindered by the current lack of design principles and predictability. Quantitative structure-property relationships have emerged as a useful technique to tie properties of interest to the molecular structure across various applications but have not been widely explored in supramolecular material problems. Herein we describe and share an accessible, visual programming, classification framework which we have applied to the prediction of solubility of amino acid-functionalised napthalenes and rylenes. We show through a proof-of-concept dataset for amino acid-functionalised naphthalenes and rylenes that solubility can be effectively classified with sufficient high-quality training data, and highlight the need for standardised characterisation procedures in the field of supramolecular chemistry if more complex and informative quantitative structure-property relationships are to be developed. Such models could ultimately afford the tailored design of functional materials.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13618208/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149076898","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":"Pockets to products: a data-driven approach for classification of monoterpene synthases.","authors":"Cathal Ó Raghallaigh, Nigel S Scrutton, Sam Hay","doi":"10.1039/d6dd00312e","DOIUrl":"10.1039/d6dd00312e","url":null,"abstract":"<p><p>Monoterpene synthases (mTSs) are a large family of enzymes, which have promising industrial applications, yet remain difficult to engineer due to complex and poorly understood sequence-function relationships. Here, we present a structure-based machine learning (ML) framework that accurately predicts whether a mTS is likely to produce linalool or limonene, chosen as canonical examples of linear and cyclic monoterpene products, respectively. Our approach identifies active site properties, which are used as ML features, enabling functional predictions that go beyond sequence alone. As residue positioning is important for monoterpene synthesis, we created an algorithm to identify structurally conserved residues in the active sites of mTSs, identifying new and existing motifs essential for both general catalysis and specific cyclisation steps. This integrated workflow thus provides insights into terpene synthases that were previously inaccessible and may offer a generalizable strategy for probing and engineering other poorly understood enzyme families. Future work could see this approach used to guide the rational design of mTSs and other hard-to-engineer enzyme families.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":"4573-4589"},"PeriodicalIF":7.1,"publicationDate":"2026-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13613485/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149066458","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":"Real-space chemistry on quantum computers: a fault-tolerant algorithm with adaptive grids and a transcorrelated extension.","authors":"César Feniou, Christopher Cherfan, Julien Zylberman, Baptiste Claudon, Jean-Philip Piquemal, Emmanuel Giner","doi":"10.1039/d6dd00002a","DOIUrl":"https://doi.org/10.1039/d6dd00002a","url":null,"abstract":"<p><p>First-quantized, real-space formulations of quantum chemistry on quantum computers are appealing: qubit count scales logarithmically with spatial resolution, and Coulomb operators achieve quadratic instead of quartic computational scaling of two-electron interactions. However, existing schemes employ uniform discretizations, so the resolution required to capture electron-nuclear cusps in high-density regions oversamples low-density regions, wasting computational resources. We address this by deploying non-uniform, molecule-adaptive grids that concentrate points where electronic density is high. Using Voronoi partitions of these grids, the molecular Hamiltonian is expressed in a Hermitian form and in a transcorrelated, isospectral form that eliminates Coulomb singularities and yields cusp-free eigenfunctions. Both formulations slot naturally into quantum eigenvalue solvers: Hermitian Quantum Phase Estimation (QPE) and the recent generalised Quantum Eigenvalue Estimation (QEVE) protocol for its non-Hermitian, transcorrelated counterpart. Numerical validation on benchmark systems confirms that this <i>ab initio</i> framework paves a promising route to ground-state chemistry on quantum hardware.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13651879/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149170472","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":"Predicting molecular structure from sparse NMR data, proof-of-concept using a graph neural network.","authors":"Ben Honoré, Calvin Yiu, Jose Napolitano-Farina, Dave Russell, Iuni Margaret Laura Trist, Ruth Dooley, Craig P Butts","doi":"10.1039/d6dd00189k","DOIUrl":"10.1039/d6dd00189k","url":null,"abstract":"<p><p>We demonstrate that graph-based neural networks are capable of molecular structure elucidation of organic molecules directly from experimentally measurable NMR spectroscopic properties, simulated by DFT. Such a system is a holy grail for researchers in chemical science for molecules in the solution state. Our graph-based transformer neural network architecture is based on 'inverting' IMPRESSION-G2 (which outputs predicted NMR properties based on input molecular structures) to instead output a predicted molecular structure in one shot from input computed NMR properties. The inclusion of scalar (<i>J</i>) coupling input NMR features is shown to be critical to this approach, correctly assigning ∼99.9% of bonds and correctly predicting >95% of molecules in one-shot when all theoretical <i>J</i> values are included (the 'Everything' model). When simulated NMR data are limited to only the experimentally realistic 1H and 13C NMR parameters ('Experimentally measurable simulated parameters' model), the success rate drops substantially to ∼10%. This can be improved by using multi-shot predictions combined with confidence metrics and systematic searching of heteroatom bonding structures to deliver 78.3% success rate on molecules with <2 nitrogens and <2 oxygens and 32.4% overall for the testing set used. This proof-of-concept demonstrates the potential to map spectroscopic properties directly onto molecular structure properties using graph neural networks and highlights key sensitivities to spectroscopic data sparsity and heteroatom degeneracy when taking this approach.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":" ","pages":""},"PeriodicalIF":7.1,"publicationDate":"2026-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13647041/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149139381","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":"Binary interaction parameters of a million mixtures for PCP-SAFT using deep learning.","authors":"Carl Hemprich, Gernot Bauer, Gustavo Chaparro, André Bardow, Philipp Rehner","doi":"10.1039/d6dd00195e","DOIUrl":"10.1039/d6dd00195e","url":null,"abstract":"<p><p>Vapor-liquid equilibria of mixtures are essential for designing processes in the chemical, pharmaceutical, and energy industries. In the absence of experimental data, early conceptual process design relies on prediction methods such as molecular equations of state. Among these, PCP-SAFT has proven to be a robust theory for modeling phase behavior. Recent regression efforts yielded high-quality PCP-SAFT parameters for 1842 pure substances and 7861 of their binary mixtures. However, the mixtures with regressed binary interaction parameters only cover a minor part of the binary mixture space spanned by the 1842 pure substances that comprises nearly 1.7 million mixtures. To fill this data gap, we propose a deep learning model that predicts binary interaction parameters of non-cross-associating mixtures. We employ a multi-layer perceptron and compare its performance for different mixture representations. Our findings indicate that a combination of PCP-SAFT parameters and structural information predicts binary interaction parameters most accurately. Introducing physical hard constraints into the model provides additional soundness without affecting its generalization performance. The resulting hard-constrained, transformer-based model is evaluated on a hold-out test set of 841 mixtures. We demonstrate that the predicted binary interaction parameters halve the median deviation in bubble/dew point and relative volatility predictions compared to the benchmark of not using the binary interaction parameter. Substantial improvements are observed across almost all combinations of chemical families in the test set. As a final result, the proposed model predicts binary interaction parameters of 1 026 523 mixtures, vastly extending the existing database.</p>","PeriodicalId":72816,"journal":{"name":"Digital discovery","volume":"5 10","pages":"4554-4572"},"PeriodicalIF":7.1,"publicationDate":"2026-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13667907/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149198096","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}