Finding optimal reaction conditions is crucial for chemical synthesis in the pharmaceutical and chemical industries. However, due to the vast chemical space, conducting experiments for all the possible combinations is impractical. Thus, quantitative structure-activity relationship (QSAR) models have been widely used to predict product yields, but evaluating all combinations is still computationally intensive. In this work, we demonstrate the use of Digital Annealer Unit (DAU) can tackle these large-scale optimization problems more efficiently. Two types of models are developed and tested on high-throughput experimentation (HTE) and Reaxys datasets. Our results suggest that the performance of models is comparable to classical machine learning (ML) methods (i.e., Random Forest and Multilayer Perceptron (MLP)), while the inference time of our models requires only seconds with a DAU. In active learning and autonomous reaction condition design, our model shows improvement for reaction yield prediction by incorporating new data, meaning that it can potentially be used in iterative processes. Our method can also accelerate the screening of billions of reaction conditions, achieving speeds millions of times faster than traditional computing units in identifying superior conditions.

In this work, a concept for an open chemistry knowledge base was developed to integrate chemical research results into a collaboratively usable platform. To achieve this, we enhanced Semantic MediaWiki (SMW) to support the collection and structured summary of chemical data contained in publications. We implemented tools for capturing chemical structures in machine-readable formats and designed data forms along with a data model to ensure standardized input and organization of research results. These enhancements allow for effective data comparison and contextual analysis within an expandable Wiki environment. The use of the platform was specifically demonstrated by organizing and comparing research in the area of “CO2 reduction in homogeneous photocatalytic systems,” showcasing its potential to significantly enhance the collaborative collection of research outcomes.Scientific contributionThis work shows ways to collaboratively collect and manage subject-specific knowledge in the domain of chemistry via an open database. By integrating cheminformatic tools into Semantic Mediawiki, an established technology for building knowledge databases is made systematically usable for the chemical community. The integration of chemistry-specific workflows and forms allows the mapping of data from current research with links to the original sources. This work is intended to show how gaps in the information system of scientists can be closed without having to use commercial systems.

Virtual Screening (VS) of large compound libraries using Artificial Intelligence (AI) models is a highly effective approach for early drug discovery. Data splitting is crucial for benchmarking the performance of such AI models. Traditional random data splits often result in structurally similar molecules in both training and test sets, which conflict with the reality of VS libraries that typically contain structurally diverse compounds. To tackle this challenge, scaffold split, which groups molecules by shared core structure, and Butina clustering, which clusters molecules by chemotypes, have long been used. However, we show that these methods still introduce high similarities between training and test sets, leading to overestimated model performance. Our study examined four representative AI models across 60 NCI-60 datasets, each comprising approximately 33,000–54,000 molecules tested on different cancer cell lines. Each dataset was split in four ways: random, scaffold, Butina clustering and the more realistic Uniform Manifold Approximation and Projection (UMAP) clustering. Using Linear Regression, Random Forest, Transformer-CNN, and GEM, we trained a total of 8400 models and evaluated under four splitting methods. These comprehensive results show that UMAP split provides more challenging and realistic benchmarks for model evaluation, followed by Butina splits, then scaffold splits and closely after random splits. Consequently, we recommend using UMAP splits instead of overly optimistic Butina splits and especially scaffold splits for molecular property prediction, including VS. Lastly, we illustrate how misaligned ROC AUC is with VS goals, despite its common use. The code and datasets for reproducibility are available at https://github.com/Rong830/UMAP_split_for_VS and archived in https://zenodo.org/records/14736486. Scientific contribution This work advances the field by introducing UMAP clustering as a robust splitting method for molecular datasets, improving over traditional methods like Butina clustering and especially scaffold splits. It offers a new evaluation framework to benchmark AI models under more realistic conditions, fostering progress in molecular property prediction. The findings also show how inappropriate the use of ROC AUC for virtual screening (VS) continues to be, despite its popularity, emphasizing the need for context-specific evaluation metrics.

Atom mapping involves identifying the correspondence between individual atoms in reactant molecules and their counterparts in product molecules. This process is crucial for gaining deeper insight into reaction mechanisms, such as defining reaction templates and determining which chemical bonds are formed or broken during a reaction. However, reliable atom mapping data are often limited or incomplete within chemical databases, rendering manual annotation impractical for large-scale datasets. To address this limitation, we propose the Symmetry-Aware Multitask Atom Mapping Network (SAMMNet), a model designed to automatically infer atom correspondences by incorporating an auxiliary self-supervised task during training. SAMMNet employs molecular graph representations and leverages graph neural networks to capture both general and task-specific features, enabling enhanced predictive performance. Our experimental results demonstrate that the multitask learning framework, coupled with symmetry-aware atom mapping, improves accuracy and robustness in atom mapping predictions. This makes our method a promising advancement for computational chemistry and related fields.

Electronic Lab Notebooks (ELNs) have become indispensable tools for modern research laboratories, facilitating data management, collaboration, and documentation of scientific experiments. However, the proliferation of diverse ELN platforms poses challenges for researchers who need to seamlessly exchange data between different systems. In this paper, we present ELNdataBridge, a novel server-based solution designed to address this challenge by providing a flexible adapter for interfacing and synchronising data between disparate ELN platforms. ELNdataBridge leverages Python APIs to interact with the underlying data structures of various ELN systems, enabling smooth transfer of information between them. The system offers a user-friendly interface that allows researchers to map and configure the transfer of single values and entry types between different ELNs, thereby facilitating interoperability and data exchange. The suitability and efficiency of the developed software was shown by a first demonstrator, enabling the exchange of data from Chemotion ELN and Herbie, and therewith the connection of information with a focus on chemistry and materials sciences.Scientific contribution: To the best of our knowledge, a method enabling the interoperable exchange of information between different ELNs, as described here, has not yet been reported. Given the increasing number of scientists using ELNs and their reliance on discipline-specific platforms, this work proposes a solution to overcome the current limitations related to ELN interoperability.

Efficient and reliable prediction of molecular properties, such as water solubility, hydration free energy, lipophilicity, and quantum mechanical properties, is essential for rational compound design in the chemical and pharmaceutical industries. While Graph Neural Networks (GNNs) have significantly advanced molecular property prediction tasks, their high memory footprint, computational demands, and inference latency are often overlooked. These challenges hinder the deployment of property prediction models on resource-constrained devices such as smartphones and IoT devices. Therefore, optimizing storage, reducing resource consumption, and improving inference speed are crucial. This paper presents a systematic approach to molecular networks by integrating GNN models with the DoReFa-Net quantization algorithm. The proposed method aims to enhance computational efficiency while maintaining predictive performance, enabling lightweight yet effective models suitable for molecular task. The study investigates the impact of different bitwidth quantization levels on model performance, using metrics such as RMSE and MAE. Results show that, for physical chemistry datasets, the effectiveness of quantization is highly dependent on the model architecture. Notably, the quantum mechanical dipole moment task maintains strong performance up to 8-bit precision, achieving similar or slightly better results. However, extreme quantization, particularly at 2-bit precision, severely degrades performance, highlighting the limitations of aggressive compression.

This work presents the LabIMotion extension for the Chemotion Electronic Lab Notebook (ELN), expanding its capabilities from organic chemistry to support interdisciplinary research and enabling the description of workflows. LabIMotion enhances documentation by introducing customizable components structured across three levels—Elements, Segments, and Datasets—enabling flexible, hierarchical organization and reuse of data. Through the integration of links to ontologies, the extension ensures precise, machine-readable data, promoting interoperability and adherence to FAIR principles. The extension features an intuitive, user-friendly interface that allows researchers to easily create new ELN content by leveraging a set of customizable, generic methods. Scientists can set up new data fields, can link data fields, or establish workflows, and the extension translates those needs directly into usable functionality at their command. Through this high degree of flexibility, a wide range of specific research needs can be met. The LabIMotion Hub plays a crucial role in distributing and updating components, fostering standardization, and enabling collaborative development within scientific communities. These advancements significantly improve the ELN's adaptability, usability, and relevance across various research disciplines.Scientific contributionThis work demonstrates how research data management systems can be designed to support discipline-specific requirements in chemistry research while offering a high flexibility and interoperability to deal with interdisciplinary work. The developed software, LabIMotion, offers a versatile approach for integrating novel research aspects into a research data environment, fostering bottom-up processes for defining schemas and standardizing scientific workflows. In particular, the software’s support for community-driven extensions, combined with a clear definition of content and its assignment to ontology terms, provides unique advantages for creating adaptable tools suited to the complexities of the scientific environment.

Organic anion transporting polypeptides (OATPs) are membrane transporters crucial for drug uptake and distribution in the human body. OATPs can mediate drug-drug interactions (DDIs) in which the interaction of one drug with an OATP impairs the uptake of another drug, resulting in potentially fatal pharmacological effects. Predicting OATP-mediated DDIs is challenging, due to limited information on OATP inhibition mechanisms and inconsistent experimental OATP inhibition data across different studies. This study introduces Heterogeneous OATP-Ligand Interaction Graph Neural Network (HOLIgraph), a novel computational model that integrates molecular modeling with a graph neural network to enhance the prediction of drug-induced OATP inhibition. By combining ligand (i.e., drug) molecular features with protein-ligand interaction data from rigorous docking simulations, HOLIgraph outperforms traditional DDI prediction models which rely solely on ligand molecular features. HOLIgraph achieved a median balanced accuracy of over 90 percent when predicting inhibitors for OATP1B1, significantly outperforming purely ligand-based models. Beyond improving inhibition prediction, the data used to train HOLIgraph can enable the characterization of protein residues involved in inhibitory drug-OATP interactions. We identified certain OATP1B1 residues that preferentially interact with inhibitors, including I46 and K49. We anticipate such interaction information will be valuable to future structural and mechanistic investigations of OATP1B1.

Assay interference caused by small organic compounds continues to pose formidable challenges to early drug discovery. Various computational methods have been developed to identify compounds likely to cause assay interference. However, due to the scarcity of data available for model development, the predictive accuracy and applicability of these approaches are limited. In this work, we present E-GuARD, a novel framework seeking to address data scarcity and imbalance by integrating self-distillation, active learning, and expert-guided molecular generation. E-GuARD iteratively enriches the training data with interference-relevant molecules, resulting in quantitative structure-interference relationship (QSIR) models with superior performance. We demonstrate the utility of E-GuARD with the examples of four high-quality data sets on thiol reactivity, redox reactivity, nanoluciferase inhibition, and firefly luciferase inhibition. Our models reached MCC values of up to 0.47 for these data sets, with two-fold or higher improvements in enrichment factors compared to models trained without E-GuARD data augmentation. These results highlight the potential of E-GuARD as a scalable solution to mitigating assay interference in early drug discovery.Scientific contributionWe present E-GuARD, an innovative framework that combines iterative self-distillation with guided molecular augmentation to enhance the predictive performance of QSAR models. By allowing models to learn from newly generated, informative compounds through iterations, E-GuARD facilitates the understanding of underrepresented structural patterns and improves performance on unseen data. When applied across different interference mechanisms, E-GuARD consistently outperformed standard approaches. E-GuARD establishes the foundation for further research into dynamic data enrichment and more robust molecular modeling.

Representing molecules as graphs is a natural approach for capturing their structural information, with atoms depicted as nodes and bonds as edges. Although graph-based similarity calculation approaches, such as the graph edit distance, have been proposed for calculating molecular similarity, these approaches are nondeterministic polynomial (NP)-hard and thus computationally infeasible for routine use, unlike fingerprint-based methods. To address this limitation, we developed GESim, an ultrafast graph-based method for calculating molecular similarity on the basis of von Neumann graph entropy. GESim enables molecular similarity calculations by considering entire molecular graphs, and evaluations using two benchmarks for molecular similarity suggest that GESim has the ability to differentiate between highly similar molecules, even in cases where other methods fail to effectively distinguish their similarity. GESim is provided as an open-source package on GitHub at https://github.com/LazyShion/GESim.