Could graph neural networks learn better molecular representation for drug discovery? A comparison study of descriptor-based and graph-based models

Multimodal fused deep learning for drug property prediction: Integrating chemical language and molecular graph

Accurately predicting molecular properties is a challenging but essential task in drug discovery. Recently, many mono-modal deep learning methods have been successfully applied to molecular property prediction. However, mono-modal learning is inherently limited as it relies solely on a single modality of molecular representation, which restricts a comprehensive understanding of drug molecules. To overcome the limitations, we propose a multimodal fused deep learning (MMFDL) model to leverage information from different molecular representations. Specifically, we construct a triple-modal learning model by employing Transformer-Encoder, Bidirectional Gated Recurrent Unit (BiGRU), and graph convolutional network (GCN) to process three modalities of information from chemical language and molecular graph: SMILES-encoded vectors, ECFP fingerprints, and molecular graphs, respectively. We evaluate the proposed triple-modal model using five fusion approaches on six molecule datasets, including Delaney, Llinas2020, Lipophilicity, SAMPL, BACE, and pKa from DataWarrior. The results show that the MMFDL model achieves the highest Pearson coefficients, and stable distribution of Pearson coefficients in the random splitting test, outperforming mono-modal models in accuracy and reliability. Furthermore, we validate the generalization ability of our model in the prediction of binding constants for protein-ligand complex molecules, and assess the resilience capability against noise. Through analysis of feature distributions in chemical space and the assigned contribution of each modal model, we demonstrate that the MMFDL model shows the ability to acquire complementary information by using proper models and suitable fusion approaches. By leveraging diverse sources of bioinformatics information, multimodal deep learning models hold the potential for successful drug discovery.

Curvature-enhanced graph convolutional network for biomolecular interaction prediction

Geometric deep learning has demonstrated a great potential in non-Euclidean data analysis. The incorporation of geometric insights into learning architecture is vital to its success. Here we propose a curvature-enhanced graph convolutional network (CGCN) for biomolecular interaction prediction. Our CGCN employs Ollivier-Ricci curvature (ORC) to characterize network local geometric properties and enhance the learning capability of GCNs. More specifically, ORCs are evaluated based on the local topology from node neighborhoods, and further incorporated into the weight function for the feature aggregation in message-passing procedure. Our CGCN model is extensively validated on fourteen real-world bimolecular interaction networks and analyzed in details using a series of well-designed simulated data. It has been found that our CGCN can achieve the state-of-the-art results. It outperforms all existing models, as far as we know, in thirteen out of the fourteen real-world datasets and ranks as the second in the rest one. The results from the simulated data show that our CGCN model is superior to the traditional GCN models regardless of the positive-to-negative-curvature ratios, network densities, and network sizes (when larger than 500).

Metal-Organic Frameworks through the Lens of Artificial Intelligence: A Comprehensive Review

Metal-organic frameworks (MOFs) are a class of hybrid porous materials that have gained prominence as a noteworthy material with varied applications. Currently, MOFs are in extensive use, particularly in the realms of energy and catalysis. The synthesis of these materials poses considerable challenges, and their computational analysis is notably intricate due to their complex structure and versatile applications in the field of material science. Density functional theory (DFT) has helped researchers in understanding reactions and mechanisms, but it is costly and time-consuming and requires bigger systems to perform these calculations. Machine learning (ML) techniques were adopted in order to overcome these problems by implementing ML in material data sets for synthesis, structure, and property predictions of MOFs. These predictions are fast, efficient, and accurate and do not require heavy computing. In this review, we discuss ML models used in MOF and their incorporation with artificial intelligence (AI) in structure and property predictions. The advantage of AI in this field would accelerate research, particularly in synthesizing novel MOFs with multiple properties and applications oriented with minimum information.

Bitter peptide prediction using graph neural networks

Bitter taste is an unpleasant taste modality that affects food consumption. Bitter peptides are generated during enzymatic processes that produce functional, bioactive protein hydrolysates or during the aging process of fermented products such as cheese, soybean protein, and wine. Understanding the underlying peptide sequences responsible for bitter taste can pave the way for more efficient identification of these peptides. This paper presents BitterPep-GCN, a feature-agnostic graph convolution network for bitter peptide prediction. The graph-based model learns the embedding of amino acids in the bitter peptide sequences and uses mixed pooling for bitter classification. BitterPep-GCN was benchmarked using BTP640, a publicly available bitter peptide dataset. The latent peptide embeddings generated by the trained model were used to analyze the activity of sequence motifs responsible for the bitter taste of the peptides. Particularly, we calculated the activity for individual amino acids and dipeptide, tripeptide, and tetrapeptide sequence motifs present in the peptides. Our analyses pinpoint specific amino acids, such as F, G, P, and R, as well as sequence motifs, notably tripeptide and tetrapeptide motifs containing FF, as key bitter signatures in peptides. This work not only provides a new predictor of bitter taste for a more efficient identification of bitter peptides in various food products but also gives a hint into the molecular basis of bitterness.Scientific ContributionOur work provides the first application of Graph Neural Networks for the prediction of peptide bitter taste. The best-developed model, BitterPep-GCN, learns the embedding of amino acids in the bitter peptide sequences and uses mixed pooling for bitter classification. The embeddings were used to analyze the sequence motifs responsible for the bitter taste.

Large-scale deep learning identifies the antiviral potential of PKI-179 and MTI-31 against coronaviruses

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) has led to the global pandemic of Coronavirus Disease (2019) (COVID-19), underscoring the urgency for effective antiviral drugs. Despite the development of different vaccination strategies, the search for specific antiviral compounds remains crucial. Here, we combine machine learning (ML) techniques with in vitro validation to efficiently identify potential antiviral compounds. We overcome the limited amount of SARS-CoV-2 data available for ML using various techniques, supplemented with data from diverse biomedical assays, which enables end-to-end training of a deep neural network architecture. We use its predictions to identify and prioritize compounds for in vitro testing. Two top-hit compounds, PKI-179 and MTI-31, originally identified as Pi3K-mTORC1/2 pathway inhibitors, exhibit significant antiviral activity against SARS-CoV-2 at low micromolar doses. Notably, both compounds outperform the well-known mTOR inhibitor rapamycin. Furthermore, PKI-179 and MTI-31 demonstrate broad-spectrum antiviral activity against SARS-CoV-2 variants of concern and other coronaviruses. In a physiologically relevant model, both compounds show antiviral effects in primary human airway epithelial (HAE) cultures derived from healthy donors cultured in an air-liquid interface (ALI). This study highlights the potential of ML combined with in vitro testing to expedite drug discovery, emphasizing the adaptability of AI-driven approaches across different viruses, thereby contributing to pandemic preparedness.

FGTN: Fragment-based graph transformer network for predicting reproductive toxicity

Reproductive toxicity is one of the important issues in chemical safety. Traditional laboratory testing methods are costly and time-consuming with raised ethical issues. Only a few in silico models have been reported to predict human reproductive toxicity, but none of them make full use of the topological information of compounds. In addition, most existing atom-based graph neural network methods focus on attributing model predictions to individual nodes or edges rather than chemically meaningful fragments or substructures. In current studies, we develop a novel fragment-based graph transformer network (FGTN) approach to generate the QSAR model of human reproductive toxicity by considering internal topological structure information of compounds. In the FGTN model, the compound is represented by a graph architecture using fragments to be nodes and bonds linking two fragments to be edges. A super molecule-level node is further proposed to connect all fragment nodes by undirected edges, obtaining global molecular features from fragment embeddings. The FGTN model achieved an accuracy (ACC) of 0.861 and an area under the receiver operating characteristic curve (AUC) value of 0.914 on nonredundant blind tests, outperforming traditional fingerprint-based machine learning models and atom-based GCN model. The FGTN model can attribute toxic predictions to fragments, generating specific structural alerts for the positive compound. Moreover, FGTN may also have the capability to distinguish various chemical isomers. We believe that FGTN can be used as a reliable and effective tool for human reproductive toxicity prediction in contribution to the advancement of chemical safety assessment.

A Novel Rational PROTACs Design and Validation via AI-Driven Drug Design Approach

The rational design of novel drug candidates presents a formidable challenge in modern drug discovery. Proteolysis-targeting chimeras (PROTACs) drug design is particularly demanding due to their limited crystal structure availability and design of a viable small molecule to bridge the protein of interest (POI) and ubiquitin-protein ligase (E3). An integrated approach that combines superimposition techniques and deep neural networks is demonstrated in this study to leverage the power of deep learning and structural biology to generate structurally diverse molecules with enhanced binding affinities. The superimposition technique ensures the congruence of initial and new protein-ligand pairs, which are evaluated via subsequent comprehensive screening using the root-mean-square deviation (RMSD), binding free energy (BFE), and buried solvent-accessible surface area (SASA). The final candidates are subjected to the incorporation of molecular dynamics (MD) and free energy perturbation (FEP) simulations to provide a quantitative evaluation of relative binding energies, reinforcing the efficacy and reliability of the generated molecules. The outcomes of the generated novel PROTACs molecules exhibit comparable structural attributes while demonstrating superior binding affinities within the binding pockets when contrasted with those of the established cocrystal ternary complexes. To enhance the generalizability of the workflow, we chose the ternary structure of the cellular inhibitor of apoptosis protein 1 (cIAP1) and Bruton's Tyrosine Kinase (BTK) for validating the chemical properties generated from the processes. The new linker molecules additionally showed superior affinity from the simulations. In summary, this methodology serves as an effective workflow to align computational predictions with current limitations, thereby introducing a novel paradigm in AI-driven drug design.

MolPipeline: A Python Package for Processing Molecules with RDKit in Scikit-learn

The open-source package scikit-learn provides various machine learning algorithms and data processing tools, including the Pipeline class, which allows users to prepend custom data transformation steps to the machine learning model. We introduce the MolPipeline package, which extends this concept to cheminformatics by wrapping standard RDKit functionality, such as reading and writing SMILES strings or calculating molecular descriptors from a molecule object. We aimed to build an easy-to-use Python package to create completely automated end-to-end pipelines that scale to large data sets. Particular emphasis was put on handling erroneous instances, where resolution would require manual intervention in default pipelines. MolPipeline provides the building blocks to enable seamless integration of common cheminformatics tasks within scikit-learn's pipeline framework, such as scaffold splits and molecular standardization, making pipeline building easily adaptable to diverse project requirements.

Integrated Transfer Learning and Multitask Learning Strategies to Construct Graph Neural Network Models for Predicting Bioaccumulation Parameters of Chemicals

Accurate prediction of parameters related to the environmental exposure of chemicals is crucial for the sound management of chemicals. However, the lack of large data sets for training models may result in poor prediction accuracy and robustness. Herein, integrated transfer learning (TL) and multitask learning (MTL) was proposed for constructing a graph neural network (GNN) model (abbreviated as TL-MTL-GNN model) using n-octanol/water partition coefficients as a source domain. The TL-MTL-GNN model was trained to predict three bioaccumulation parameters based on enlarged data sets that cover 2496 compounds with at least one bioaccumulation parameter. Results show that the TL-MTL-GNN model outperformed single-task GNN models with and without the TL, as well as conventional machine learning models trained with molecular descriptors or fingerprints. Applicability domains were characterized by a state-of-the-art structure-activity landscape-based (abbreviated as ADSAL) methodology. The TL-MTL-GNN model coupled with the optimal ADSAL was employed to predict bioaccumulation parameters for around 60,000 chemicals, with more than 13,000 compounds identified as bioaccumulative chemicals. The high predictive accuracy and robustness of the TL-MTL-GNN model demonstrate the feasibility of integrating the TL and MTL strategy in modeling small-sized data sets. The strategy holds significant potential for addressing small data challenges in modeling environmental chemicals.

Application of Virtual Drug Study to New Drug Research and Development: Challenges and Opportunity

In recent years, virtual drug study, as an emerging research strategy, has become increasingly important in guiding and promoting new drug research and development. Researchers can integrate a variety of technical methods to improve the efficiency of all phases of new drug research and development, including the use of artificial intelligence, modeling and simulation for target identification, compound screening and pharmacokinetic characteristics evaluation, and the application of clinical trial simulation to carry out clinical research. This paper aims to elaborate on the application of virtual drug study in the key stages of new drug research and development and discuss the opportunities and challenges it faces in supporting new drug research and development.

3D physiologically-informed deep learning for drug discovery of a novel vascular endothelial growth factor receptor-2 (VEGFR2)

Angiogenesis is an essential process in tumorigenesis, tumor invasion, and metastasis, and is an intriguing pathway for drug discovery. Targeting vascular endothelial growth factor receptor 2 (VEGFR2) to inhibit tumor angiogenic pathways has been widely explored and adopted in clinical practice. However, most drugs, such as the Food and Drug Administration -approved drug axitinib (ATC code: L01EK01), have considerable side effects and limited tolerability. Therefore, there is an urgent need for the development of novel VEGFR2 inhibitors. In this study, we propose a novel strategy to design potential candidates targeting VEGFR2 using three-dimensional (3D) deep learning and structural modeling methods. A geometric-enhanced molecular representation learning method (GEM) model employing a graph neural network (GNN) as its underlying predictive algorithm was used to predict the activity of the candidates. In the structural modeling method, flexible docking was performed to screen data with high affinity and explore the mechanism of the inhibitors. Small -molecule compounds with consistently improved properties were identified based on the intersection of the scores obtained from both methods. Candidates identified using the GEM-GNN model were selected for in silico modeling using molecular dynamics simulations to further validate their efficacy. The GEM-GNN model enabled the identification of candidate compounds with potentially more favorable properties than the existing drug, axitinib, while achieving higher efficacy.

Research on prediction of human oral bioavailability of drugs based on improved deep forest

Human oral bioavailability is a crucial factor in drug discovery. In recent years, researchers have constructed a variety of different prediction models. However, given the limited size of human oral bioavailability data sets, the challenge of making accurate predictions with small sample sizes has become a critical issue in the field. The deep forest model, with its adaptively determinable number of cascade levels, can perform exceptionally well even on small-scale data. However, the original deep forest suffers unbalanced multi-grained scanning process and premature stopping of cascade forest training. In this paper, we propose a human oral bioavailability predict method based on an improved deep forest, called balanced multi-grained scanning mapping cascade forest (bgmc-forest). Firstly, the mordred descriptor method is selected to feature extraction, then enhanced features are obtained by the improved balanced multi-grained scanning, which solves the problem of missing features at both ends. And finally, the prediction results are obtained by feature mapping cascaded forests, which is based on principal component analysis and cascade forests, ensures the effectiveness of the cascade forest. The superiority of the model constructed in this paper is demonstrated through comparative experiments, while the effectiveness of the improved module is verified through ablation experiments. Finally the decision-making process of the model is explained by the shapley additive explanations interpretation algorithm.

QupKake: Integrating Machine Learning and Quantum Chemistry for Micro-pKa Predictions

Accurate prediction of micro-pKa values is crucial for understanding and modulating the acidity and basicity of organic molecules, with applications in drug discovery, materials science, and environmental chemistry. This work introduces QupKake, a novel method that combines graph neural network models with semiempirical quantum mechanical (QM) features to achieve exceptional accuracy and generalization in micro-pKa prediction. QupKake outperforms state-of-the-art models on a variety of benchmark data sets, with root-mean-square errors between 0.5 and 0.8 pKa units on five external test sets. Feature importance analysis reveals the crucial role of QM features in both the reaction site enumeration and micro-pKa prediction models. QupKake represents a significant advancement in micro-pKa prediction, offering a powerful tool for various applications in chemistry and beyond.

BartSmiles: Generative Masked Language Models for Molecular Representations

We discover a robust self-supervised strategy tailored toward molecular representations for generative masked language models through a series of tailored, in-depth ablations. Using this pretraining strategy, we train BARTSmiles, a BART-like model with an order of magnitude more compute than previous self-supervised molecular representations. In-depth evaluations show that BARTSmiles consistently outperforms other self-supervised representations across classification, regression, and generation tasks, setting a new state-of-the-art on eight tasks. We then show that when applied to the molecular domain, the BART objective learns representations that implicitly encode our downstream tasks of interest. For example, by selecting seven neurons from a frozen BARTSmiles, we can obtain a model having performance within two percentage points of the full fine-tuned model on task Clintox. Lastly, we show that standard attribution interpretability methods, when applied to BARTSmiles, highlight certain substructures that chemists use to explain specific properties of molecules. The code and pretrained model are publicly available.

Classification models for predicting the bioactivity of pan-TRK inhibitors and SAR analysis

Tropomyosin receptor kinases (TRKs) are important broad-spectrum anticancer targets. The oncogenic rearrangement of the NTRK gene disrupts the extracellular structural domain and epitopes for therapeutic antibodies, making small-molecule inhibitors essential for treating NTRK fusion-driven tumors. In this work, several algorithms were used to construct descriptor-based and nondescriptor-based models, and the models were evaluated by outer 10-fold cross-validation. To find a model with good generalization ability, the dataset was partitioned by random and cluster-splitting methods to construct in- and cross-domain models, respectively. Among the 48 models built, the model with the combination of the deep neural network (DNN) algorithm and extended connectivity fingerprints 4 (ECFP4) descriptors achieved excellent performance in both dataset divisions. The results indicate that the DNN algorithm has a strong generalization prediction ability, and the richness of features plays a vital role in predicting unknown spatial molecules. Additionally, we combined the clustering results and decision tree models of fingerprint descriptors to perform structure-activity relationship analysis. It was found that nitrogen-containing aromatic heterocyclic and benzo heterocyclic structures play a crucial role in enhancing the activity of TRK inhibitors.

Meta Learning With Graph Attention Networks for Low-Data Drug Discovery

Finding candidate molecules with favorable pharmacological activity, low toxicity, and proper pharmacokinetic properties is an important task in drug discovery. Deep neural networks have made impressive progress in accelerating and improving drug discovery. However, these techniques rely on a large amount of label data to form accurate predictions of molecular properties. At each stage of the drug discovery pipeline, usually, only a few biological data of candidate molecules and derivatives are available, indicating that the application of deep neural networks for low-data drug discovery is still a formidable challenge. Here, we propose a meta learning architecture with graph attention network, Meta-GAT, to predict molecular properties in low-data drug discovery. The GAT captures the local effects of atomic groups at the atom level through the triple attentional mechanism and implicitly captures the interactions between different atomic groups at the molecular level. GAT is used to perceive molecular chemical environment and connectivity, thereby effectively reducing sample complexity. Meta-GAT further develops a meta learning strategy based on bilevel optimization, which transfers meta knowledge from other attribute prediction tasks to low-data target tasks. In summary, our work demonstrates how meta learning can reduce the amount of data required to make meaningful predictions of molecules in low-data scenarios. Meta learning is likely to become the new learning paradigm in low-data drug discovery. The source code is publicly available at: https://github.com/lol88/Meta-GAT.

Stochastic machine learning via sigma profiles to build a digital chemical space

This work establishes a different paradigm on digital molecular spaces and their efficient navigation by exploiting sigma profiles. To do so, the remarkable capability of Gaussian processes (GPs), a type of stochastic machine learning model, to correlate and predict physicochemical properties from sigma profiles is demonstrated, outperforming state-of-the-art neural networks previously published. The amount of chemical information encoded in sigma profiles eases the learning burden of machine learning models, permitting the training of GPs on small datasets which, due to their negligible computational cost and ease of implementation, are ideal models to be combined with optimization tools such as gradient search or Bayesian optimization (BO). Gradient search is used to efficiently navigate the sigma profile digital space, quickly converging to local extrema of target physicochemical properties. While this requires the availability of pretrained GP models on existing datasets, such limitations are eliminated with the implementation of BO, which can find global extrema with a limited number of iterations. A remarkable example of this is that of BO toward boiling temperature optimization. Holding no knowledge of chemistry except for the sigma profile and boiling temperature of carbon monoxide (the worst possible initial guess), BO finds the global maximum of the available boiling temperature dataset (over 1,000 molecules encompassing more than 40 families of organic and inorganic compounds) in just 15 iterations (i.e., 15 property measurements), cementing sigma profiles as a powerful digital chemical space for molecular optimization and discovery, particularly when little to no experimental data is initially available.

Developing a Semi-Supervised Approach Using a PU-Learning-Based Data Augmentation Strategy for Multitarget Drug Discovery

Multifactorial diseases demand therapeutics that can modulate multiple targets for enhanced safety and efficacy, yet the clinical approval of multitarget drugs remains rare. The integration of machine learning (ML) and deep learning (DL) in drug discovery has revolutionized virtual screening. This study investigates the synergy between ML/DL methodologies, molecular representations, and data augmentation strategies. Notably, we found that SVM can match or even surpass the performance of state-of-the-art DL methods. However, conventional data augmentation often involves a trade-off between the true positive rate and false positive rate. To address this, we introduce Negative-Augmented PU-bagging (NAPU-bagging) SVM, a novel semi-supervised learning framework. By leveraging ensemble SVM classifiers trained on resampled bags containing positive, negative, and unlabeled data, our approach is capable of managing false positive rates while maintaining high recall rates. We applied this method to the identification of multitarget-directed ligands (MTDLs), where high recall rates are critical for compiling a list of interaction candidate compounds. Case studies demonstrate that NAPU-bagging SVM can identify structurally novel MTDL hits for ALK-EGFR with favorable docking scores and binding modes, as well as pan-agonists for dopamine receptors. The NAPU-bagging SVM methodology should serve as a promising avenue to virtual screening, especially for the discovery of MTDLs.

Advancements in Ligand-Based Virtual Screening through the Synergistic Integration of Graph Neural Networks and Expert-Crafted Descriptors

The fusion of traditional chemical descriptors with Graph Neural Networks (GNNs) offers a compelling strategy for enhancing ligand-based virtual screening methodologies. A comprehensive evaluation revealed that the benefits derived from this integrative strategy vary significantly among different GNNs. Specifically, while GCN and SchNet demonstrate pronounced improvements by incorporating descriptors, SphereNet exhibits only marginal enhancement. Intriguingly, despite SphereNet's modest gain, all three models-GCN, SchNet, and SphereNet-achieve comparable performance levels when leveraging this combination strategy. This observation underscores a pivotal insight: sophisticated GNN architectures may be substituted with simpler counterparts without sacrificing efficacy, provided that they are augmented with descriptors. Furthermore, our analysis reveals a set of expert-crafted descriptors' robustness in scaffold-split scenarios, frequently outperforming the combined GNN-descriptor models. Given the critical importance of scaffold splitting in accurately mimicking real-world drug discovery contexts, this finding accentuates an imperative for GNN researchers to innovate models that can adeptly navigate and predict within such frameworks. Our work not only validates the potential of integrating descriptors with GNNs in advancing ligand-based virtual screening but also illuminates pathways for future enhancements in model development and application. Our implementation can be found at https://github.com/meilerlab/gnn-descriptor.

Evaluating Molecular Complexity with Open-Source Machine Learning Approaches to Predict Process Mass Intensity

The application of green chemistry is critical for cultivating environmental responsibility and sustainable practices in pharmaceutical manufacturing. Process mass intensity (PMI) is a key metric that quantifies the resource efficiency of a manufacturing process, but determining what constitutes a successful PMI of a specific molecule is challenging. A recent approach correlated molecular features to a crowdsourced definition of molecular complexity to determine PMI targets. While recent machine learning tools show promise in predicting molecular complexity, a more extensive application could significantly optimize manufacturing processes. To this end, we refine and expand upon the SMART-PMI tool by Sheridan et al. to create an open-source model and application. Our solution emphasizes explainability and parsimony to facilitate a nuanced understanding of prediction and ensure informed decision-making. The resulting model uses four descriptors-the heteroatom count, stereocenter count, unique topological torsion, and connectivity index chi4n-to compute molecular complexity with a comparable 82.6% predictive accuracy and 0.349 RMSE. We develop a corresponding app that takes in structured data files (SDF) to rapidly quantify molecular complexity and provide a PMI target that can be used to drive process development activities. By integrating machine learning explainability and open-source accessibility, we provide flexible tools to advance the field of green chemistry and sustainable pharmaceutical manufacturing.

Prediction and Interpretation Microglia Cytotoxicity by Machine Learning

Ameliorating microglia-mediated neuroinflammation is a crucial strategy in developing new drugs for neurodegenerative diseases. Plant compounds are an important screening target for the discovery of drugs for the treatment of neurodegenerative diseases. However, due to the spatial complexity of phytochemicals, it becomes particularly important to evaluate the effectiveness of compounds while avoiding the mixing of cytotoxic substances in the early stages of compound screening. Traditional high-throughput screening methods suffer from high cost and low efficiency. A computational model based on machine learning provides a novel avenue for cytotoxicity determination. In this study, a microglia cytotoxicity classifier was developed using a machine learning approach. First, we proposed a data splitting strategy based on the molecule murcko generic scaffold, under this condition, three machine learning approaches were coupled with three kinds of molecular representation methods to construct microglia cytotoxicity classifier, which were then compared and assessed by the predictive accuracy, balanced accuracy, F1-score, and Matthews Correlation Coefficient. Then, the recursive feature elimination integrated with support vector machine (RFE-SVC) dimension reduction method was introduced to molecular fingerprints with high dimensions to further improve the model performance. Among all the microglial cytotoxicity classifiers, the SVM coupled with ECFP4 fingerprint after feature selection (ECFP4-RFE-SVM) obtained the most accurate classification for the test set (ACC of 0.99, BA of 0.99, F1-score of 0.99, MCC of 0.97). Finally, the Shapley additive explanations (SHAP) method was used in interpreting the microglia cytotoxicity classifier and key substructure smart identified as structural alerts. Experimental results show that ECFP4-RFE-SVM have reliable classification capability for microglia cytotoxicity, and SHAP can not only provide a rational explanation for microglia cytotoxicity predictions, but also offer a guideline for subsequent molecular cytotoxicity modifications.

Synergizing Chemical Structures and Bioassay Descriptions for Enhanced Molecular Property Prediction in Drug Discovery

The precise prediction of molecular properties can greatly accelerate the development of new drugs. However, in silico molecular property prediction approaches have been limited so far to assays for which large amounts of data are available. In this study, we develop a new computational approach leveraging both the textual description of the assay of interest and the chemical structure of target compounds. By combining these two sources of information via self-supervised learning, our tool can provide accurate predictions for assays where no measurements are available. Remarkably, our approach achieves state-of-the-art performance on the FS-Mol benchmark for zero-shot prediction, outperforming a wide variety of deep learning approaches. Additionally, we demonstrate how our tool can be used for tailoring screening libraries for the assay of interest, showing promising performance in a retrospective case study on a high-throughput screening campaign. By accelerating the early identification of active molecules in drug discovery and development, this method has the potential to streamline the identification of novel therapeutics.

Highly Accurate and Explainable Predictions of Small-Molecule Antioxidants for Eight In Vitro Assays Simultaneously through an Alternating Multitask Learning Strategy

Small molecule antioxidants can inhibit or retard oxidation reactions and protect against free radical damage to cells, thus playing a key role in food, cosmetics, pharmaceuticals, the environment, as well as materials. Experimentally driven antioxidant discovery is a major paradigm, and computationally assisted antioxidants are rarely reported. In this study, a functional-group-based alternating multitask self-supervised molecular representation learning method is proposed to simultaneously predict the antioxidant activities of small molecules for eight commonly used in vitro antioxidant assays. Extensive evaluation results reveal that compared with the baseline models, the multitask FG-BERT model achieves the best overall predictive performance, with the highest average F1, BA, ROC-AUC, and PRC-AUC values of 0.860, 0.880, 0.954, and 0.937 for the test sets, respectively. The Y-scrambling testing results further demonstrate that such a deep learning model was not constructed by accident and that it has reliable predictive capabilities. Additionally, the excellent interpretability of the multitask FG-BERT model makes it easy to identify key structural fragments/groups that contribute significantly to the antioxidant effect of a given molecule. Finally, an online antioxidant activity prediction platform called AOP (freely available at https://aop.idruglab.cn/) and its local version were developed based on the high-quality multitask FG-BERT model for experts and nonexperts in the field. We anticipate that it will contribute to the discovery of novel small-molecule antioxidants.

Reducing overconfident errors in molecular property classification using Posterior Network

Deep-learning-based classification models are increasingly used for predicting molecular properties in drug development. However, traditional classification models using the Softmax function often give overconfident mispredictions for out-of-distribution samples, highlighting a critical lack of accurate uncertainty estimation. Such limitations can result in substantial costs and should be avoided during drug development. Inspired by advances in evidential deep learning and Posterior Network, we replaced the Softmax function with a normalizing flow to enhance the uncertainty estimation ability of the model in molecular property classification. The proposed strategy was evaluated across diverse scenarios, including simulated experiments based on a synthetic dataset, ADMET predictions, and ligand-based virtual screening. The results demonstrate that compared with the vanilla model, the proposed strategy effectively alleviates the problem of giving overconfident but incorrect predictions. Our findings support the promising application of evidential deep learning in drug development and offer a valuable framework for further research.

Prediction of blood-brain barrier permeability using machine learning approaches based on various molecular representation

The assessment of compound blood-brain barrier (BBB) permeability poses a significant challenge in the discovery of drugs targeting the central nervous system. Conventional experimental approaches to measure BBB permeability are labor-intensive, cost-ineffective, and time-consuming. In this study, we constructed six machine learning classification models by combining various machine learning algorithms and molecular representations. The model based on ExtraTree algorithm and random partitioning strategy obtains the best prediction result, with AUC value of 0.932±0.004 and balanced accuracy (BA) of 0.837±0.010 for the test set. We employed the SHAP method to identify important features associated with BBB permeability. In addition, matched molecular pair (MMP) analysis and representative substructure derivation method were utilized to uncover the transformation rules and distinctive structural features of BBB permeable compounds. The machine learning models proposed in this work can serve as an effective tool for assessing BBB permeability in the drug discovery for central nervous system disease.

Unsupervised manifold embedding to encode molecular quantum information for supervised learning of chemical data

Molecular representation is critical in chemical machine learning. It governs the complexity of model development and the fulfillment of training data to avoid either over- or under-fitting. As electronic structures and associated attributes are the root cause for molecular interactions and their manifested properties, we have sought to examine the local electron information on a molecular manifold to understand and predict molecular interactions. Our efforts led to the development of a lower-dimensional representation of a molecular manifold, Manifold Embedding of Molecular Surface (MEMS), to embody surface electronic quantities. By treating a molecular surface as a manifold and computing its embeddings, the embedded electronic attributes retain the chemical intuition of molecular interactions. MEMS can be further featurized as input for chemical learning. Our solubility prediction with MEMS demonstrated the feasibility of both shallow and deep learning by neural networks, suggesting that MEMS is expressive and robust against dimensionality reduction.

Cancer drug sensitivity estimation using modular deep Graph Neural Networks

Computational drug sensitivity models have the potential to improve therapeutic outcomes by identifying targeted drugs components that are tailored to the transcriptomic profile of a given primary tumor. The SMILES representation of molecules that is used by state-of-the-art drug-sensitivity models is not conducive for neural networks to generalize to new drugs, in part because the distance between atoms does not generally correspond to the distance between their representation in the SMILES strings. Graph-attention networks, on the other hand, are high-capacity models that require large training-data volumes which are not available for drug-sensitivity estimation. We develop a modular drug-sensitivity graph-attentional neural network. The modular architecture allows us to separately pre-train the graph encoder and graph-attentional pooling layer on related tasks for which more data are available. We observe that this model outperforms reference models for the use cases of precision oncology and drug discovery; in particular, it is better able to predict the specific interaction between drug and cell line that is not explained by the general cytotoxicity of the drug and the overall survivability of the cell line. The complete source code is available at https://zenodo.org/doi/10.5281/zenodo.8020945. All experiments are based on the publicly available GDSC data.

Machine learning models to predict ligand binding affinity for the orexin 1 receptor

The orexin 1 receptor (OX1R) is a G-protein coupled receptor that regulates a variety of physiological processes through interactions with the neuropeptides orexin A and B. Selective OX1R antagonists exhibit therapeutic effects in preclinical models of several behavioral disorders, including drug seeking and overeating. However, currently there are no selective OX1R antagonists approved for clinical use, fueling demand for novel compounds that act at this target. In this study, we meticulously curated a dataset comprising over 1300 OX1R ligands using a stringent filter and criteria cascade. Subsequently, we developed highly predictive quantitative structure-activity relationship (QSAR) models employing the optimized hyper-parameters for the random forest machine learning algorithm and twelve 2D molecular descriptors selected by recursive feature elimination with a 5-fold cross-validation process. The predictive capacity of the QSAR model was further assessed using an external test set and enrichment study, confirming its high predictivity. The practical applicability of our final QSAR model was demonstrated through virtual screening of the DrugBank database. This revealed two FDA-approved drugs (isavuconazole and cabozantinib) as potential OX1R ligands, confirmed by radiolabeled OX1R binding assays. To our best knowledge, this study represents the first report of highly predictive QSAR models on a large comprehensive dataset of diverse OX1R ligands, which should prove useful for the discovery and design of new compounds targeting this receptor.

PointGAT: A Quantum Chemical Property Prediction Model Integrating Graph Attention and 3D Geometry

Predicting quantum chemical properties is a fundamental challenge for computational chemistry. While the development of graph neural networks has advanced molecular representation learning and property prediction, their performance could be further enhanced by incorporating three-dimensional (3D) structural geometry into two-dimensional (2D) molecular graph representation. In this study, we introduce the PointGAT model for quantum molecular property prediction, which integrates 3D molecular coordinates with graph-attention modeling. Comparison with other current models in molecular prediction tasks showed that PointGAT could provide higher predictive accuracy in various benchmark data sets from MoleculeNet, including ESOL, FreeSolv, Lipop, HIV, and 6 out of 12 tasks of the QM9 data set. To further examine PointGAT prediction of quantum mechanical (QM) energies, we constructed a C10 data set comprising 11,841 charged and chiral carbocation intermediates with QM energies calculated at the DM21/6-31G*//B3LYP/6-31G* levels. Notably, PointGAT achieved an R2 value of 0.950 and an MAE of 1.616 kcal/mol, outperforming even the best-performing graph neural network model with a reduction of 0.216 kcal/mol in MAE and an improvement of 0.050 in R2. Additional ablation studies indicated that incorporating molecular geometry into the model resulted in markedly higher predictive accuracy, reducing the MAE value from 1.802 to 1.616 kcal/mol. Moreover, visualization of PointGAT atomic attention weights suggested its predictions were interpretable. Findings in this study support the application of PointGAT as a powerful and versatile tool for quantum chemical property prediction that can facilitate high-accuracy modeling for fundamental exploration of chemical space as well as drug design and molecular engineering.

Knowledge mapping of graph neural networks for drug discovery: a bibliometric and visualized analysis

Introduction

In recent years, graph neural network has been extensively applied to drug discovery research. Although researchers have made significant progress in this field, there is less research on bibliometrics. The purpose of this study is to conduct a comprehensive bibliometric analysis of graph neural network applications in drug discovery in order to identify current research hotspots and trends, as well as serve as a reference for future research.

Methods

Publications from 2017 to 2023 about the application of graph neural network in drug discovery were collected from the Web of Science Core Collection. Bibliometrix, VOSviewer, and Citespace were mainly used for bibliometric studies.

Results and Discussion

In this paper, a total of 652 papers from 48 countries/regions were included. Research interest in this field is continuously increasing. China and the United States have a significant advantage in terms of funding, the number of publications, and collaborations with other institutions and countries. Although some cooperation networks have been formed in this field, extensive worldwide cooperation still needs to be strengthened. The results of the keyword analysis clarified that graph neural network has primarily been applied to drug-target interaction, drug repurposing, and drug-drug interaction, while graph convolutional neural network and its related optimization methods are currently the core algorithms in this field. Data availability and ethical supervision, balancing computing resources, and developing novel graph neural network models with better interpretability are the key technical issues currently faced. This paper analyzes the current state, hot spots, and trends of graph neural network applications in drug discovery through bibliometric approaches, as well as the current issues and challenges in this field. These findings provide researchers with valuable insights on the current status and future directions of this field.

AAGCN: a graph convolutional neural network with adaptive feature and topology learning

In recent years, there has been a growing prevalence of deep learning in various domains, owing to advancements in information technology and computing power. Graph neural network methods within deep learning have shown remarkable capabilities in processing graph-structured data, such as social networks and traffic networks. As a result, they have garnered significant attention from researchers.However, real-world data often face challenges like data sparsity and missing labels, which can hinder the performance and generalization ability of graph convolutional neural networks. To overcome these challenges, our research aims to effectively extract the hidden features and topological information of graph convolutional neural networks. We propose an innovative model called Adaptive Feature and Topology Graph Convolutional Neural Network (AAGCN). By incorporating an adaptive layer, our model preprocesses the data and integrates the hidden features and topological information with the original data's features and structure. These fused features are then utilized in the convolutional layer for training, significantly enhancing the expressive power of graph convolutional neural networks.To evaluate the effectiveness of the adaptive layer in the AAGCN model, we conducted node classification experiments on real datasets. The results validate its ability to address data sparsity and improve the classification performance of graph convolutional neural networks.In conclusion, our research primarily focuses on addressing data sparsity and missing labels in graph convolutional neural networks. The proposed AAGCN model, which incorporates an adaptive layer, effectively extracts hidden features and topological information, thereby enhancing the expressive power and classification performance of these networks.

Development of a Novel In Silico Classification Model to Assess Reactive Metabolite Formation in the Cysteine Trapping Assay and Investigation of Important Substructures

Predicting whether a compound can cause drug-induced liver injury (DILI) is difficult due to the complexity of drug mechanism. The cysteine trapping assay is a method for detecting reactive metabolites that bind to microsomes covalently. However, it is cumbersome to use 35S isotope-labeled cysteine for this assay. Therefore, we constructed an in silico classification model for predicting a positive/negative outcome in the cysteine trapping assay. We collected 475 compounds (436 in-house compounds and 39 publicly available drugs) based on experimental data performed in this study, and the composition of the results showed 248 positives and 227 negatives. Using a Message Passing Neural Network (MPNN) and Random Forest (RF) with extended connectivity fingerprint (ECFP) 4, we built machine learning models to predict the covalent binding risk of compounds. In the time-split dataset, AUC-ROC of MPNN and RF were 0.625 and 0.559 in the hold-out test, restrictively. This result suggests that the MPNN model has a higher predictivity than RF in the time-split dataset. Hence, we conclude that the in silico MPNN classification model for the cysteine trapping assay has a better predictive power. Furthermore, most of the substructures that contributed positively to the cysteine trapping assay were consistent with previous results.

Distinguishing Molecular Properties of OAT, OATP, and MRP Drug Substrates by Machine Learning

The movement of organic anionic drugs across cell membranes is partly governed by interactions with SLC and ABC transporters in the intestine, liver, kidney, blood-brain barrier, placenta, breast, and other tissues. Major transporters involved include organic anion transporters (OATs, SLC22 family), organic anion transporting polypeptides (OATPs, SLCO family), and multidrug resistance proteins (MRPs, ABCC family). However, the sets of molecular properties of drugs that are necessary for interactions with OATs (OAT1, OAT3) vs. OATPs (OATP1B1, OATP1B3) vs. MRPs (MRP2, MRP4) are not well-understood. Defining these molecular properties is necessary for a better understanding of drug and metabolite handling across the gut-liver-kidney axis, gut-brain axis, and other multi-organ axes. It is also useful for tissue targeting of small molecule drugs and predicting drug-drug interactions and drug-metabolite interactions. Here, we curated a database of drugs shown to interact with these transporters in vitro and used chemoinformatic approaches to describe their molecular properties. We then sought to define sets of molecular properties that distinguish drugs interacting with OATs, OATPs, and MRPs in binary classifications using machine learning and artificial intelligence approaches. We identified sets of key molecular properties (e.g., rotatable bond count, lipophilicity, number of ringed structures) for classifying OATs vs. MRPs and OATs vs. OATPs. However, sets of molecular properties differentiating OATP vs. MRP substrates were less evident, as drugs interacting with MRP2 and MRP4 do not form a tight group owing to differing hydrophobicity and molecular complexity for interactions with the two transporters. If the results also hold for endogenous metabolites, they may deepen our knowledge of organ crosstalk, as described in the Remote Sensing and Signaling Theory. The results also provide a molecular basis for understanding how small organic molecules differentially interact with OATs, OATPs, and MRPs.

Machine Learning Assisted Hit Prioritization for High Throughput Screening in Drug Discovery

Efficient prioritization of bioactive compounds from high throughput screening campaigns is a fundamental challenge for accelerating drug development efforts. In this study, we present the first data-driven approach to simultaneously detect assay interferents and prioritize true bioactive compounds. By analyzing the learning dynamics during training of a gradient boosting model on noisy high throughput screening data using a novel formulation of sample influence, we are able to distinguish between compounds exhibiting the desired biological response and those producing assay artifacts. Therefore, our method enables false positive and true positive detection without relying on prior screens or assay interference mechanisms, making it applicable to any high throughput screening campaign. We demonstrate that our approach consistently excludes assay interferents with different mechanisms and prioritizes biologically relevant compounds more efficiently than all tested baselines, including a retrospective case study simulating its use in a real drug discovery campaign. Finally, our tool is extremely computationally efficient, requiring less than 30 s per assay on low-resource hardware. As such, our findings show that our method is an ideal addition to existing false positive detection tools and can be used to guide further pharmacological optimization after high throughput screening campaigns.

Enhancing Multi-species Liver Microsomal Stability Prediction through Artificial Intelligence

Liver microsomal stability, a crucial aspect of metabolic stability, significantly impacts practical drug discovery. However, current models for predicting liver microsomal stability are based on limited molecular information from a single species. To address this limitation, we constructed the largest public database of compounds from three common species: human, rat, and mouse. Subsequently, we developed a series of classification models using both traditional descriptor-based and classic graph-based machine learning (ML) algorithms. Remarkably, the best-performing models for the three species achieved Matthews correlation coefficients (MCCs) of 0.616, 0.603, and 0.574, respectively, on the test set. Furthermore, through the construction of consensus models based on these individual models, we have demonstrated their superior predictive performance in comparison with the existing models of the same type. To explore the similarities and differences in the properties of liver microsomal stability among multispecies molecules, we conducted preliminary interpretative explorations using the Shapley additive explanations (SHAP) and atom heatmap approaches for the models and misclassified molecules. Additionally, we further investigated representative structural modifications and substructures that decrease the liver microsomal stability in different species using the matched molecule pair analysis (MMPA) method and substructure extraction techniques. The established prediction models, along with insightful interpretation information regarding liver microsomal stability, will significantly contribute to enhancing the efficiency of exploring practical drugs for development.

Classification of substances by health hazard using deep neural networks and molecular electron densities

In this paper we present a method that allows leveraging 3D electron density information to train a deep neural network pipeline to segment regions of high, medium and low electronegativity and classify substances as health hazardous or non-hazardous. We show that this can be used for use-cases such as cosmetics and food products. For this purpose, we first generate 3D electron density cubes using semiempirical molecular calculations for a custom European Chemicals Agency (ECHA) subset consisting of substances labelled as hazardous and non-hazardous for cosmetic usage. Together with their 3-class electronegativity maps we train a modified 3D-UNet with electron density cubes to segment reactive sites in molecules and classify substances with an accuracy of 78.1%. We perform the same process on a custom food dataset (CompFood) consisting of hazardous and non-hazardous substances compiled from European Food Safety Authority (EFSA) OpenFoodTox, Food and Drug Administration (FDA) Generally Recognized as Safe (GRAS) and FooDB datasets to achieve a classification accuracy of 64.1%. Our results show that 3D electron densities and particularly masked electron densities, calculated by taking a product of original electron densities and regions of high and low electronegativity can be used to classify molecules for different use-cases and thus serve not only to guide safe-by-design product development but also aid in regulatory decisions. SCIENTIFIC CONTRIBUTION: We aim to contribute to the diverse 3D molecular representations used for training machine learning algorithms by showing that a deep learning network can be trained on 3D electron density representation of molecules. This approach has previously not been used to train machine learning models and it allows utilization of the true spatial domain of the molecule for prediction of properties such as their suitability for usage in cosmetics and food products and in future, to other molecular properties. The data and code used for training is accessible at https://github.com/s-singh-ivv/eDen-Substances .

HyperPCM: Robust Task-Conditioned Modeling of Drug-Target Interactions

A central problem in drug discovery is to identify the interactions between drug-like compounds and protein targets. Over the past few decades, various quantitative structure-activity relationship (QSAR) and proteo-chemometric (PCM) approaches have been developed to model and predict these interactions. While QSAR approaches solely utilize representations of the drug compound, PCM methods incorporate both representations of the protein target and the drug compound, enabling them to achieve above-chance predictive accuracy on previously unseen protein targets. Both QSAR and PCM approaches have recently been improved by machine learning and deep neural networks, that allow the development of drug-target interaction prediction models from measurement data. However, deep neural networks typically require large amounts of training data and cannot robustly adapt to new tasks, such as predicting interaction for unseen protein targets at inference time. In this work, we propose to use HyperNetworks to efficiently transfer information between tasks during inference and thus to accurately predict drug-target interactions on unseen protein targets. Our HyperPCM method reaches state-of-the-art performance compared to previous methods on multiple well-known benchmarks, including Davis, DUD-E, and a ChEMBL derived data set, and particularly excels at zero-shot inference involving unseen protein targets. Our method, as well as reproducible data preparation, is available at https://github.com/ml-jku/hyper-dti.

Benchmarking of Small Molecule Feature Representations for hERG, Nav1.5, and Cav1.2 Cardiotoxicity Prediction

In the field of drug discovery, there is a substantial challenge in seeking out chemical structures that possess desirable pharmacological, toxicological, and pharmacokinetic properties. Complications arise when drugs interfere with the functioning of cardiac ion channels, leading to serious cardiovascular consequences. The discontinuation and removal of numerous approved drugs from the market or at late development stages in the pipeline due to such inhibitory effects further highlight the urgency of addressing this issue. Consequently, the early prediction of potential blockers targeting cardiac ion channels during the drug discovery process is of paramount importance. This study introduces a deep learning framework that computationally determines the cardiotoxicity associated with the voltage-gated potassium channel (hERG), the voltage-gated calcium channel (Cav1.2), and the voltage-gated sodium channel (Nav1.5) for drug candidates. The predictive capabilities of three feature representations─molecular fingerprints, descriptors, and graph-based numerical representations─are rigorously benchmarked. Additionally, a novel training and evaluation data set framework is presented, enabling predictive model training of drug off-target cardiotoxicity using a comprehensive and large curated data set covering these three cardiac ion channels. To facilitate these predictions, a robust and comprehensive small molecule cardiotoxicity prediction tool named CToxPred has been developed. It is made available as open source under the permissive MIT license at https://github.com/issararab/CToxPred.

Advancing material property prediction: using physics-informed machine learning models for viscosity

In materials science, accurately computing properties like viscosity, melting point, and glass transition temperatures solely through physics-based models is challenging. Data-driven machine learning (ML) also poses challenges in constructing ML models, especially in the material science domain where data is limited. To address this, we integrate physics-informed descriptors from molecular dynamics (MD) simulations to enhance the accuracy and interpretability of ML models. Our current study focuses on accurately predicting viscosity in liquid systems using MD descriptors. In this work, we curated a comprehensive dataset of over 4000 small organic molecules' viscosities from scientific literature, publications, and online databases. This dataset enabled us to develop quantitative structure-property relationships (QSPR) consisting of descriptor-based and graph neural network models to predict temperature-dependent viscosities for a wide range of viscosities. The QSPR models reveal that including MD descriptors improves the prediction of experimental viscosities, particularly at the small data set scale of fewer than a thousand data points. Furthermore, feature importance tools reveal that intermolecular interactions captured by MD descriptors are most important for viscosity predictions. Finally, the QSPR models can accurately capture the inverse relationship between viscosity and temperature for six battery-relevant solvents, some of which were not included in the original data set. Our research highlights the effectiveness of incorporating MD descriptors into QSPR models, which leads to improved accuracy for properties that are difficult to predict when using physics-based models alone or when limited data is available.

Informative Training Data for Efficient Property Prediction in Metal-Organic Frameworks by Active Learning

In recent data-driven approaches to material discovery, scenarios where target quantities are expensive to compute and measure are often overlooked. In such cases, it becomes imperative to construct a training set that includes the most diverse, representative, and informative samples. Here, a novel regression tree-based active learning algorithm is employed for such a purpose. It is applied to predict the band gap and adsorption properties of metal-organic frameworks (MOFs), a novel class of materials that results from the virtually infinite combinations of their building units. Simpler and low dimensional descriptors, such as those based on stoichiometric and geometric properties, are used to compute the feature space for this model owing to their ability to better represent MOFs in the low data regime. The partitions given by a regression tree constructed on the labeled part of the data set are used to select new samples to be added to the training set, thereby limiting its size while maximizing the prediction quality. Tests on the QMOF, hMOF, and dMOF data sets reveal that our method constructs small training data sets to learn regression models that predict the target properties more efficiently than existing active learning approaches, and with lower variance. Specifically, our active learning approach is highly beneficial when labels are unevenly distributed in the descriptor space and when the label distribution is imbalanced, which is often the case for real world data. The regions defined by the tree help in revealing patterns in the data, thereby offering a unique tool to efficiently analyze complex structure-property relationships in materials and accelerate materials discovery.

CatNet: Sequence-based deep learning with cross-attention mechanism for identifying endocrine-disrupting chemicals

Endocrine-disrupting chemicals (EDCs) pose significant environmental and health risks due to their potential to interfere with nuclear receptors (NRs), key regulators of physiological processes. Despite the evident risks, the majority of existing research narrows its focus on the interaction between compounds and the individual NR target, neglecting a comprehensive assessment across the entire NR family. In response, this study assembled a comprehensive human NR dataset, capturing 49,244 interactions between 35,467 unique compounds and 42 NRs. We introduced a cross-attention network framework, "CatNet", innovatively integrating compound and protein representations through cross-attention mechanisms. The results showed that CatNet model achieved excellent performance with an area under the receiver operating characteristic curve (AUCROC) = 0.916 on the test set, and exhibited reliable generalization on unseen compound-NR pairs. A distinguishing feature of our research is its capacity to expand to novel targets. Beyond its predictive accuracy, CatNet offers a valuable mechanistic perspective on compound-NR interactions through feature visualization. Augmenting the utility of our research, we have also developed a graphical user interface, empowering researchers to predict chemical binding to diverse NRs. Our model enables the prediction of human NR-related EDCs and shows the potential to identify EDCs related to other targets.

Improving chemical reaction yield prediction using pre-trained graph neural networks

Graph neural networks (GNNs) have proven to be effective in the prediction of chemical reaction yields. However, their performance tends to deteriorate when they are trained using an insufficient training dataset in terms of quantity or diversity. A promising solution to alleviate this issue is to pre-train a GNN on a large-scale molecular database. In this study, we investigate the effectiveness of GNN pre-training in chemical reaction yield prediction. We present a novel GNN pre-training method for performance improvement.Given a molecular database consisting of a large number of molecules, we calculate molecular descriptors for each molecule and reduce the dimensionality of these descriptors by applying principal component analysis. We define a pre-text task by assigning a vector of principal component scores as the pseudo-label to each molecule in the database. A GNN is then pre-trained to perform the pre-text task of predicting the pseudo-label for the input molecule. For chemical reaction yield prediction, a prediction model is initialized using the pre-trained GNN and then fine-tuned with the training dataset containing chemical reactions and their yields. We demonstrate the effectiveness of the proposed method through experimental evaluation on benchmark datasets.

Genetic Algorithm-Based Receptor Ligand: A Genetic Algorithm-Guided Generative Model to Boost the Novelty and Drug-Likeness of Molecules in a Sampling Chemical Space

Deep learning-based de novo molecular design has recently gained significant attention. While numerous DL-based generative models have been successfully developed for designing novel compounds, the majority of the generated molecules lack sufficiently novel scaffolds or high drug-like profiles. The aforementioned issues may not be fully captured by commonly used metrics for the assessment of molecular generative models, such as novelty, diversity, and quantitative estimation of the drug-likeness score. To address these limitations, we proposed a genetic algorithm-guided generative model called GARel (genetic algorithm-based receptor-ligand interaction generator), a novel framework for training a DL-based generative model to produce drug-like molecules with novel scaffolds. To efficiently train the GARel model, we utilized dense net to update the parameters based on molecules with novel scaffolds and drug-like features. To demonstrate the capability of the GARel model, we used it to design inhibitors for three targets: AA2AR, EGFR, and SARS-Cov2. The results indicate that GARel-generated molecules feature more diverse and novel scaffolds and possess more desirable physicochemical properties and favorable docking scores. Compared with other generative models, GARel makes significant progress in balancing novelty and drug-likeness, providing a promising direction for the further development of DL-based de novo design methodology with potential impacts on drug discovery.

Accurate Electronic and Optical Properties of Organic Doublet Radicals Using Machine Learned Range-Separated Functionals

Luminescent organic semiconducting doublet-spin radicals are unique and emergent optical materials because their fluorescent quantum yields (Φfl) are not compromised by the spin-flipping intersystem crossing (ISC) into a dark high-spin state. The multiconfigurational nature of these radicals challenges their electronic structure calculations in the framework of single-reference density functional theory (DFT) and introduces room for method improvement. In the present study, we extended our earlier development of ML-ωPBE [J. Phys. Chem. Lett., 2021, 12, 9516-9524], a range-separated hybrid (RSH) exchange-correlation (XC) functional constructed using the stacked ensemble machine learning (SEML) algorithm, from closed-shell organic semiconducting molecules to doublet-spin organic semiconducting radicals. We assessed its performance for a new test set of 64 doublet-spin radicals from five categories while placing all previously compiled 3926 closed-shell molecules in the new training set. Interestingly, ML-ωPBE agrees with the nonempirical OT-ωPBE functional regarding the prediction of the molecule-dependent range-separation parameter (ω), with a small mean absolute error (MAE) of 0.0197 a0-1, but saves the computational cost by 2.46 orders of magnitude. This result demonstrates an outstanding domain adaptation capacity of ML-ωPBE for diverse organic semiconducting species. To further assess the predictive power of ML-ωPBE in experimental observables, we also applied it to evaluate absorption and fluorescence energies (Eabs and Efl) using linear-response time-dependent DFT (TDDFT), and we compared its behavior with nine popular XC functionals. For most radicals, ML-ωPBE reproduces experimental measurements of Eabs and Efl with small MAEs of 0.299 and 0.254 eV, only marginally different from those of OT-ωPBE. Our work illustrates a successful extension of the SEML framework from closed-shell molecules to doublet-spin radicals and will open the venue for calculating optical properties for organic semiconductors using single-reference TDDFT.

Novel symmetry-preserving neural network model for phylogenetic inference

Motivation

Scientists world-wide are putting together massive efforts to understand how the biodiversity that we see on Earth evolved from single-cell organisms at the origin of life and this diversification process is represented through the Tree of Life. Low sampling rates and high heterogeneity in the rate of evolution across sites and lineages produce a phenomenon denoted "long branch attraction" (LBA) in which long nonsister lineages are estimated to be sisters regardless of their true evolutionary relationship. LBA has been a pervasive problem in phylogenetic inference affecting different types of methodologies from distance-based to likelihood-based.

Results

Here, we present a novel neural network model that outperforms standard phylogenetic methods and other neural network implementations under LBA settings. Furthermore, unlike existing neural network models in phylogenetics, our model naturally accounts for the tree isomorphisms via permutation invariant functions which ultimately result in lower memory and allows the seamless extension to larger trees.

Availability and implementation

We implement our novel theory on an open-source publicly available GitHub repository: https://github.com/crsl4/nn-phylogenetics.

Generative artificial intelligence in drug discovery: basic framework, recent advances, challenges, and opportunities

There are two main ways to discover or design small drug molecules. The first involves fine-tuning existing molecules or commercially successful drugs through quantitative structure-activity relationships and virtual screening. The second approach involves generating new molecules through de novo drug design or inverse quantitative structure-activity relationship. Both methods aim to get a drug molecule with the best pharmacokinetic and pharmacodynamic profiles. However, bringing a new drug to market is an expensive and time-consuming endeavor, with the average cost being estimated at around $2.5 billion. One of the biggest challenges is screening the vast number of potential drug candidates to find one that is both safe and effective. The development of artificial intelligence in recent years has been phenomenal, ushering in a revolution in many fields. The field of pharmaceutical sciences has also significantly benefited from multiple applications of artificial intelligence, especially drug discovery projects. Artificial intelligence models are finding use in molecular property prediction, molecule generation, virtual screening, synthesis planning, repurposing, among others. Lately, generative artificial intelligence has gained popularity across domains for its ability to generate entirely new data, such as images, sentences, audios, videos, novel chemical molecules, etc. Generative artificial intelligence has also delivered promising results in drug discovery and development. This review article delves into the fundamentals and framework of various generative artificial intelligence models in the context of drug discovery via de novo drug design approach. Various basic and advanced models have been discussed, along with their recent applications. The review also explores recent examples and advances in the generative artificial intelligence approach, as well as the challenges and ongoing efforts to fully harness the potential of generative artificial intelligence in generating novel drug molecules in a faster and more affordable manner. Some clinical-level assets generated form generative artificial intelligence have also been discussed in this review to show the ever-increasing application of artificial intelligence in drug discovery through commercial partnerships.

Immunotherapy and targeted therapy for cholangiocarcinoma: Artificial intelligence research in imaging

Cholangiocarcinoma (CCA) is a highly aggressive hepatobiliary malignancy, second only to hepatocellular carcinoma in prevalence. Despite surgical treatment being the recommended method to achieve a cure, it is not viable for patients with advanced CCA. Gene sequencing and artificial intelligence (AI) have recently opened up new possibilities in CCA diagnosis, treatment, and prognosis assessment. Basic research has furthered our understanding of the tumor-immunity microenvironment and revealed targeted molecular mechanisms, resulting in immunotherapy and targeted therapy being increasingly employed in the clinic. Yet, the application of these remedies in CCA is a challenging endeavor due to the varying pathological mechanisms of different CCA types and the lack of expressed immune proteins and molecular targets in some patients. AI in medical imaging has emerged as a powerful tool in this situation, as machine learning and deep learning are able to extract intricate data from CCA lesion images while assisting clinical decision making, and ultimately improving patient prognosis. This review summarized and discussed the current immunotherapy and targeted therapy related to CCA, and the research progress of AI in this field.

Computational modeling approaches for developing a synergistic effect prediction model of estrogen agonistic activity

The concerns regarding the potential health threats caused by estrogenic endocrine-disrupting chemicals (EDCs) and their mixtures manufactured by the chemical industry are increasing worldwide. Conventional experimental tests for understanding the estrogenic activity of mixtures are expensive and time-consuming. Although non-testing methods using computational modeling approaches have been developed to reduce the number of traditional tests, they are unsuitable for predicting synergistic effects because current prediction models consider only a single chemical. Thus, the development of predictive models is essential for predicting the mixture toxicity, including chemical interactions. However, selecting suitable computational modeling approaches to develop a high-performance prediction model requires considerable time and effort. In this study, we provide a suitable computational approach to develop a predictive model for the synergistic effects of estrogenic activity. We collected datasets on mixture toxicity based on the synergistic effect of estrogen agonistic activity in binary mixtures. Using the model deviation ratio approach, we classified the labels of the binary mixtures as synergistic or non-synergistic effects. We assessed five molecular descriptors, four machine learning-based algorithms, and a deep learning-based algorithm to provide a suitable computational modeling approach. Compared with other modeling approaches, the prediction model using the deep learning-based algorithm and chemical-protein network descriptors exhibited the best performance in predicting the synergistic effects. In conclusion, we developed a new high-performance binary classification model using a deep neural network and chemical-protein network-based descriptors. The developed model will be helpful for the preliminary screening of the synergistic effects of binary mixtures during the development process of chemical products.

Is fragment-based graph a better graph-based molecular representation for drug design? A comparison study of graph-based models

Graph Neural Networks (GNNs) have gained significant traction in various sectors of AI-driven drug design. Over recent years, the integration of fragmentation concepts into GNNs has emerged as a potent strategy to augment the efficacy of molecular generative models. Nonetheless, challenges such as symmetry breaking and potential misrepresentation of intricate cycles and undefined functional groups raise questions about the superiority of fragment-based graph representation over traditional methods. In our research, we undertook a rigorous evaluation, contrasting the predictive prowess of eight models-developed using deep learning algorithms-across 12 benchmark datasets that span a range of properties. These models encompass established methods like GCN, AttentiveFP, and D-MPNN, as well as innovative fragment-based representation techniques. Our results indicate that fragment-based methodologies, notably PharmHGT, significantly improve model performance and interpretability, particularly in scenarios characterized by limited data availability. However, in situations with extensive training, fragment-based molecular graph representations may not necessarily eclipse traditional methods. In summation, we posit that the integration of fragmentation, as an avant-garde technique in drug design, harbors considerable promise for the future of AI-enhanced drug design.

Large-scale comparison of machine learning methods for profiling prediction of kinase inhibitors

Conventional machine learning (ML) and deep learning (DL) play a key role in the selectivity prediction of kinase inhibitors. A number of models based on available datasets can be used to predict the kinase profile of compounds, but there is still controversy about the advantages and disadvantages of ML and DL for such tasks. In this study, we constructed a comprehensive benchmark dataset of kinase inhibitors, involving in 141,086 unique compounds and 216,823 well-defined bioassay data points for 354 kinases. We then systematically compared the performance of 12 ML and DL methods on the kinase profiling prediction task. Extensive experimental results reveal that (1) Descriptor-based ML models generally slightly outperform fingerprint-based ML models in terms of predictive performance. RF as an ensemble learning approach displays the overall best predictive performance. (2) Single-task graph-based DL models are generally inferior to conventional descriptor- and fingerprint-based ML models, however, the corresponding multi-task models generally improves the average accuracy of kinase profile prediction. For example, the multi-task FP-GNN model outperforms the conventional descriptor- and fingerprint-based ML models with an average AUC of 0.807. (3) Fusion models based on voting and stacking methods can further improve the performance of the kinase profiling prediction task, specifically, RF::AtomPairs + FP2 + RDKitDes fusion model performs best with the highest average AUC value of 0.825 on the test sets. These findings provide useful information for guiding choices of the ML and DL methods for the kinase profiling prediction tasks. Finally, an online platform called KIPP ( https://kipp.idruglab.cn ) and python software are developed based on the best models to support the kinase profiling prediction, as well as various kinase inhibitor identification tasks including virtual screening, compound repositioning and target fishing.