Deep learning methods for molecular representation and property prediction

Transformers for Molecular Property Prediction: Lessons Learned from the Past Five Years

Molecular Property Prediction (MPP) is vital for drug discovery, crop protection, and environmental science. Over the last decades, diverse computational techniques have been developed, from using simple physical and chemical properties and molecular fingerprints in statistical models and classical machine learning to advanced deep learning approaches. In this review, we aim to distill insights from current research on employing transformer models for MPP. We analyze the currently available models and explore key questions that arise when training and fine-tuning a transformer model for MPP. These questions encompass the choice and scale of the pretraining data, optimal architecture selections, and promising pretraining objectives. Our analysis highlights areas not yet covered in current research, inviting further exploration to enhance the field's understanding. Additionally, we address the challenges in comparing different models, emphasizing the need for standardized data splitting and robust statistical analysis.

MSH-DTI: multi-graph convolution with self-supervised embedding and heterogeneous aggregation for drug-target interaction prediction

BACKGROUND

The rise of network pharmacology has led to the widespread use of network-based computational methods in predicting drug target interaction (DTI). However, existing DTI prediction models typically rely on a limited amount of data to extract drug and target features, potentially affecting the comprehensiveness and robustness of features. In addition, although multiple networks are used for DTI prediction, the integration of heterogeneous information often involves simplistic aggregation and attention mechanisms, which may impose certain limitations.

RESULTS

MSH-DTI, a deep learning model for predicting drug-target interactions, is proposed in this paper. The model uses self-supervised learning methods to obtain drug and target structure features. A Heterogeneous Interaction-enhanced Feature Fusion Module is designed for multi-graph construction, and the graph convolutional networks are used to extract node features. With the help of an attention mechanism, the model focuses on the important parts of different features for prediction. Experimental results show that the AUROC and AUPR of MSH-DTI are 0.9620 and 0.9605 respectively, outperforming other models on the DTINet dataset.

CONCLUSION

The proposed MSH-DTI is a helpful tool to discover drug-target interactions, which is also validated through case studies in predicting new DTIs.

DAS-DDI: A dual-view framework with drug association and drug structure for drug-drug interaction prediction

In drug development and clinical application, drug-drug interaction (DDI) prediction is crucial for patient safety and therapeutic efficacy. However, traditional methods for DDI prediction often overlook the structural features of drugs and the complex interrelationships between them, which affect the accuracy and interpretability of the model. In this paper, a novel dual-view DDI prediction framework, DAS-DDI is proposed. Firstly, a drug association network is constructed based on similarity information among drugs, which could provide rich context information for DDI prediction. Subsequently, a novel drug substructure extraction method is proposed, which could update the features of nodes and chemical bonds simultaneously, improving the comprehensiveness of the feature. Furthermore, an attention mechanism is employed to fuse multiple drug embeddings from different views dynamically, enhancing the discriminative ability of the model in handling multi-view data. Comparative experiments on three public datasets demonstrate the superiority of DAS-DDI compared with other state-of-the-art models under two scenarios.

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.

Enhancing Molecular Property Prediction through Task-Oriented Transfer Learning: Integrating Universal Structural Insights and Domain-Specific Knowledge

Precisely predicting molecular properties is crucial in drug discovery, but the scarcity of labeled data poses a challenge for applying deep learning methods. While large-scale self-supervised pretraining has proven an effective solution, it often neglects domain-specific knowledge. To tackle this issue, we introduce Task-Oriented Multilevel Learning based on BERT (TOML-BERT), a dual-level pretraining framework that considers both structural patterns and domain knowledge of molecules. TOML-BERT achieved state-of-the-art prediction performance on 10 pharmaceutical datasets. It has the capability to mine contextual information within molecular structures and extract domain knowledge from massive pseudo-labeled data. The dual-level pretraining accomplished significant positive transfer, with its two components making complementary contributions. Interpretive analysis elucidated that the effectiveness of the dual-level pretraining lies in the prior learning of a task-related molecular representation. Overall, TOML-BERT demonstrates the potential of combining multiple pretraining tasks to extract task-oriented knowledge, advancing molecular property prediction in drug discovery.

OLB-AC: toward optimizing ligand bioactivities through deep graph learning and activity cliffs

MOTIVATION

Deep graph learning (DGL) has been widely employed in the realm of ligand-based virtual screening. Within this field, a key hurdle is the existence of activity cliffs (ACs), where minor chemical alterations can lead to significant changes in bioactivity. In response, several DGL models have been developed to enhance ligand bioactivity prediction in the presence of ACs. Yet, there remains a largely unexplored opportunity within ACs for optimizing ligand bioactivity, making it an area ripe for further investigation.

RESULTS

We present a novel approach to simultaneously predict and optimize ligand bioactivities through DGL and ACs (OLB-AC). OLB-AC possesses the capability to optimize ligand molecules located near ACs, providing a direct reference for optimizing ligand bioactivities with the matching of original ligands. To accomplish this, a novel attentive graph reconstruction neural network and ligand optimization scheme are proposed. Attentive graph reconstruction neural network reconstructs original ligands and optimizes them through adversarial representations derived from their bioactivity prediction process. Experimental results on nine drug targets reveal that out of the 667 molecules generated through OLB-AC optimization on datasets comprising 974 low-activity, noninhibitor, or highly toxic ligands, 49 are recognized as known highly active, inhibitor, or nontoxic ligands beyond the datasets' scope. The 27 out of 49 matched molecular pairs generated by OLB-AC reveal novel transformations not present in their training sets. The adversarial representations employed for ligand optimization originate from the gradients of bioactivity predictions. Therefore, we also assess OLB-AC's prediction accuracy across 33 different bioactivity datasets. Results show that OLB-AC achieves the best Pearson correlation coefficient (r2) on 27/33 datasets, with an average improvement of 7.2%-22.9% against the state-of-the-art bioactivity prediction methods.

AVAILABILITY AND IMPLEMENTATION

The code and dataset developed in this work are available at github.com/Yueming-Yin/OLB-AC.

In silico prediction of ocular toxicity of compounds using explainable machine learning and deep learning approaches

The accurate identification of chemicals with ocular toxicity is of paramount importance in health hazard assessment. In contemporary chemical toxicology, there is a growing emphasis on refining, reducing, and replacing animal testing in safety evaluations. Therefore, the development of robust computational tools is crucial for regulatory applications. The performance of predictive models is heavily reliant on the quality and quantity of data. In this investigation, we amalgamated the most extensive dataset (4901 compounds) sourced from governmental GHS-compliant databases and literature to develop binary classification models of chemical ocular toxicity. We employed 12 molecular representations in conjunction with six machine learning algorithms and two deep learning algorithms to create a series of binary classification models. The findings indicated that the deep learning method GCN outperformed the machine learning models in cross-validation, achieving an impressive AUC of 0.915. However, the top-performing machine learning model (RF-Descriptor) demonstrated excellent performance with an AUC of 0.869 on the test set and was therefore selected as the best model. To enhance model interpretability, we conducted the SHAP method and attention weights analysis. The two approaches offered visual depictions of the relevance of key descriptors and substructures in predicting ocular toxicity of chemicals. Thus, we successfully struck a delicate balance between data quality and model interpretability, rendering our model valuable for predicting and comprehending potential ocular-toxic compounds in the early stages of drug discovery.

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.

Protein coding regions prediction by fusing DNA shape features

Exons crucial for coding are often hidden within introns, and the two tend to vary greatly in length, which results in deep learning-based protein coding region prediction methods often performing poorly when applied to more structurally complex biological genomes. DNA shape information also plays a role in revealing the underlying logic of gene expression, yet current methods ignore the influence of DNA shape features when distinguishing coding and non-coding regions. We propose a method to predict protein-coding regions using the CNNS-BRNN model, which incorporates DNA shape features and improves the model's ability to distinguish between intronic and exonic features. We use a fusion coding technique that combines DNA shape features and traditional sequence features. Experiments show that this method outperforms the baseline method in metrics such as AUC and F1 by 2.3% and 5.3%, respectively, and the fusion coding method that introduces DNA shape features has a significant improvement in model performance.

Comparing ANI-2x, ANI-1ccx neural networks, force field, and DFT methods for predicting conformational potential energy of organic molecules

In this study, the conformational potential energy surfaces of Amylmetacresol, Benzocaine, Dopamine, Betazole, and Betahistine molecules were scanned and analyzed using the neural network architecture ANI-2 × and ANI-1ccx, the force field method OPLS, and density functional theory with the exchange-correlation functional B3LYP and the basis set 6-31G(d). The ANI-1ccx and ANI-2 × methods demonstrated the highest accuracy in predicting torsional energy profiles, effectively capturing the minimum and maximum values of these profiles. Conformational potential energy values calculated by B3LYP and the OPLS force field method differ from those calculated by ANI-1ccx and ANI-2x, which account for non-bonded intramolecular interactions, since the B3LYP functional and OPLS force field weakly consider van der Waals and other intramolecular forces in torsional energy profiles. For a more comprehensive analysis, electronic parameters such as dipole moment, HOMO, and LUMO energies for different torsional angles were calculated at two levels of theory, B3LYP/6-31G(d) and ωB97X/6-31G(d). These calculations confirmed that ANI predictions are more accurate than density functional theory calculations with B3LYP functional and OPLS force field for determining potential energy surfaces. This research successfully addressed the challenges in determining conformational potential energy levels and shows how machine learning and deep neural networks offer a more accurate, cost-effective, and rapid alternative for predicting torsional energy profiles.

CollaPPI: A Collaborative Learning Framework for Predicting Protein-Protein Interactions

Exploring protein-protein interaction (PPI) is of paramount importance for elucidating the intrinsic mechanism of various biological processes. Nevertheless, experimental determination of PPI can be both time-consuming and expensive, motivating the exploration of data-driven deep learning technologies as a viable, efficient, and accurate alternative. Nonetheless, most current deep learning-based methods regarded a pair of proteins to be predicted for possible interaction as two separate entities when extracting PPI features, thus neglecting the knowledge sharing among the collaborative protein and the target protein. Aiming at the above issue, a collaborative learning framework CollaPPI was proposed in this study, where two kinds of collaboration, i.e., protein-level collaboration and task-level collaboration, were incorporated to achieve not only the knowledge-sharing between a pair of proteins, but also the complementation of such shared knowledge between biological domains closely related to PPI (i.e., protein function, and subcellular location). Evaluation results demonstrated that CollaPPI obtained superior performance compared to state-of-the-art methods on two PPI benchmarks. Besides, evaluation results of CollaPPI on the additional PPI type prediction task further proved its excellent generalization ability.

INTransformer: Data augmentation-based contrastive learning by injecting noise into transformer for molecular property prediction

Molecular property prediction plays an essential role in drug discovery for identifying the candidate molecules with target properties. Deep learning models usually require sufficient labeled data to train good prediction models. However, the size of labeled data is usually small for molecular property prediction, which brings great challenges to deep learning-based molecular property prediction methods. Furthermore, the global information of molecules is critical for predicting molecular properties. Therefore, we propose INTransformer for molecular property prediction, which is a data augmentation method via contrastive learning to alleviate the limitations of the labeled molecular data while enhancing the ability to capture global information. Specifically, INTransformer consists of two identical Transformer sub-encoders to extract the molecular representation from the original SMILES and noisy SMILES respectively, while achieving the goal of data augmentation. To reduce the influence of noise, we use contrastive learning to ensure the molecular encoding of noisy SMILES is consistent with that of the original input so that the molecular representation information can be better extracted by INTransformer. Experiments on various benchmark datasets show that INTransformer achieved competitive performance for molecular property prediction tasks compared with the baselines and state-of-the-art methods.

Exploring Novel Fentanyl Analogues Using a Graph-Based Transformer Model

The structures of fentanyl and its analogues are easy to be modified and few types have been included in database so far, which allow criminals to avoid the supervision of relevant departments. This paper introduces a molecular graph-based transformer model, which is combined with a data augmentation method based on substructure replacement to generate novel fentanyl analogues. 140,000 molecules were generated, and after a set of screening, 36,799 potential fentanyl analogues were finally obtained. We calculated the molecular properties of 36,799 potential fentanyl analogues. The results showed that the model could learn some properties of original fentanyl molecules. We compared the generated molecules from transformer model and data augmentation method based on substructure replacement with those generated by the other two molecular generation models based on deep learning, and found that the model in this paper can generate more novel potential fentanyl analogues. Finally, the findings of the paper indicate that transformer model based on molecular graph helps us explore the structure of potential fentanyl molecules as well as understand distribution of original molecules of fentanyl.

A focus on molecular representation learning for the prediction of chemical properties

Molecular representation learning (MRL) is a specialized field in which deep-learning models condense essential molecular information into a vectorized form. Whereas recent research has predominantly emphasized drug discovery and bioactivity applications, MRL holds significant potential for diverse chemical properties beyond these contexts. The recently published study by King-Smith introduces a novel application of molecular representation training and compellingly demonstrates its value in predicting molecular properties (E. King-Smith, Chem. Sci., 2024, https://doi.org/10.1039/D3SC04928K). In this focus article, we will briefly delve into MRL in chemistry and the significance of King-Smith's work within the dynamic landscape of this evolving field.

Pre-training molecular representation model with spatial geometry for property prediction

AI-enhanced bioinformatics and cheminformatics pivots on generating increasingly descriptive and generalized molecular representation. Accurate prediction of molecular properties needs a comprehensive description of molecular geometry. We design a novel Graph Isomorphic Network (GIN) based model integrating a three-level network structure with a dual-level pre-training approach that aligns the characteristics of molecules. In our Spatial Molecular Pre-training (SMPT) Model, the network can learn implicit geometric information in layers from lower to higher according to the dimension. Extensive evaluations against established baseline models validate the enhanced efficacy of SMPT, with notable accomplishments in classification tasks. These results emphasize the importance of spatial geometric information in molecular representation modeling and demonstrate the potential of SMPT as a valuable tool for property prediction.

Prediction of molecular-specific mutagenic alerts and related mechanisms of chemicals by a convolutional neural network (CNN) model based on SMILES split

Structural alerts (SAs) are essential to identify chemicals for toxicity evaluation and health risk assessment. We constructed a novel SMILES split-based deep learning model (SSDL) that was trained and verified with 5850 chemicals from the ISSSTY database and 384 external test chemicals from published papers. The training accuracy was above 0.90 and the evaluation metrics (precision, recall and F1-score) all reached 0.78 or above on both internal and external test chemicals. In this model, the molecular-specific fragment importance of chemicals was first quantified independently. Then, the SA identification method based on the importance of these fragments was statistically analyzed and verified with the ISSSTY test and external test chemicals containing one of 28 typical SAs, and most of the performances were better than that of expert rules. Furthermore, a mutagenicity mechanism prediction method was developed using 237 chemicals with four known mutagenic mechanisms based on molecular similarity calibrated by the SSDL method and fragment importance, which significantly improved accuracy in three mechanisms and had comparable accuracy in the other one compared to traditional methods. Overall, the SSDL model quantifying fragment toxicity within molecules would be a novel potentially powerful tool in the determination and visualization of molecular-specific SAs and the prediction of mutagenicity mechanisms for environmental or industrial compounds and drugs.

Calculation of solvation force in molecular dynamics simulation by deep-learning method

Electrostatic calculations are generally used in studying the thermodynamics and kinetics of biomolecules in solvent. Generally, this is performed by solving the Poisson-Boltzmann equation on a large grid system, a process known to be time consuming. In this study, we developed a deep neural network to predict the decomposed solvation free energies and forces of all atoms in a molecule. To train the network, the internal coordinates of the molecule were used as the input data, and the solvation free energies along with transformed atomic forces from the Poisson-Boltzmann equation were used as labels. Both the training and prediction tasks were accelerated on GPU. Formal tests demonstrated that our method can provide reasonable predictions for small molecules when the network is well-trained with its simulation data. This method is suitable for processing lots of snapshots of molecules in a long trajectory. Moreover, we applied this method in the molecular dynamics simulation with enhanced sampling. The calculated free energy landscape closely resembled that obtained from explicit solvent simulations.

AIRI: Predicting Retention Indices and Their Uncertainties Using Artificial Intelligence

The Kováts retention index (RI) is a quantity measured using gas chromatography and is commonly used in the identification of chemical structures. Creating libraries of observed RI values is a laborious task, so we explore the use of a deep neural network for predicting RI values from structure for standard semipolar columns. This network generated predictions with a mean absolute error of 15.1 and, in a quantification of the tail of the error distribution, a 95th percentile absolute error of 46.5. Because of the Artificial Intelligence Retention Indices (AIRI) network's accuracy, it was used to predict RI values for the NIST EI-MS spectral libraries. These RI values are used to improve chemical identification methods and the quality of the library. Estimating uncertainty is an important practical need when using prediction models. To quantify the uncertainty of our network for each individual prediction, we used the outputs of an ensemble of 8 networks to calculate a predicted standard deviation for each RI value prediction. This predicted standard deviation was corrected to follow the error between the observed and predicted RI values. The Z scores using these predicted standard deviations had a standard deviation of 1.52 and a 95th percentile absolute Z score corresponding to a mean RI value of 42.6.

A deep learning framework for predicting molecular property based on multi-type features fusion

Extracting expressive molecular features is essential for molecular property prediction. Sequence-based representation is a common representation of molecules, which ignores the structure information of molecules. While molecular graph representation has a weak ability in expressing the 3D structure. In this article, we try to make use of the advantages of different type representations simultaneously for molecular property prediction. Thus, we propose a fusion model named DLF-MFF, which integrates the multi-type molecular features. Specifically, we first extract four different types of features from molecular fingerprints, 2D molecular graph, 3D molecular graph and molecular image. Then, in order to learn molecular features individually, we use four essential deep learning frameworks, which correspond to four distinct molecular representations. The final molecular representation is created by integrating the four feature vectors and feeding them into prediction layer to predict molecular property. We compare DLF-MFF with 7 state-of-the-art methods on 6 benchmark datasets consisting of multiple molecular properties, the experimental results show that DLF-MFF achieves state-of-the-art performance on 6 benchmark datasets. Moreover, DLF-MFF is applied to identify potential anti-SARS-CoV-2 inhibitor from 2500 drugs. We predict probability of each drug being inferred as a 3CL protease inhibitor and also calculate the binding affinity scores between each drug and 3CL protease. The results show that DLF-MFF product better performance in the identification of anti-SARS-CoV-2 inhibitor. This work is expected to offer novel research perspectives for accurate prediction of molecular properties and provide valuable insights into drug repurposing for COVID-19.

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.

A deep learning and docking simulation-based virtual screening strategy enables the rapid identification of HIF-1α pathway activators from a marine natural product database

Artificial Intelligence is hailed as a cutting-edge technology for accelerating drug discovery efforts, and our goal was to validate its potential in predicting pharmacological inhibitors of EGLN1 using a deep learning-based architecture, one of its subsidiaries. Egl nine homolog 1 (EGLN1) inhibition prevents poly ubiquitination-mediated proteosomal destruction HIF-1α. The pharmacological interventions aimed at stabilizing HIF-1α have the potential to be a promising treatment option for a range of human diseases, including ischemic stroke. To unveil a novel EGLN1 inhibitor from marine natural products, a custom-based virtual screening was carried out using a Deep Convolutional Neural Network (DCNN) architecture, docking, and molecular dynamics simulation. The custom DCNN model was optimized and further employed to screen marine natural products from the CMNPD database. The docking was performed as a secondary strategy for screened hits. Molecular dynamics (MD) and molecular mechanics/generalized Born surface area (MM-GBSA) were used to analyze inhibitor binding and identify key interactions. The findings support the claim that deep learning-based virtual screening is a rapid, reliable and accurate method of identifying highly contributing drug candidates (EGLN1 inhibitors). This study demonstrates that deep learning architecture can significantly accelerate drug discovery and development, and provides a solid foundation for using (Z)-2-ethylhex-2-enedioic acid [(Z)-2-ethylhex-2-enedioic acid] as a potential EGLN1 inhibitor for treating various health complications.Communicated by Ramaswamy H. Sarma.

Deep learning algorithms applied to computational chemistry

Recently, there has been a significant increase in the use of deep learning techniques in the molecular sciences, which have shown high performance on datasets and the ability to generalize across data. However, no model has achieved perfect performance in solving all problems, and the pros and cons of each approach remain unclear to those new to the field. Therefore, this paper aims to review deep learning algorithms that have been applied to solve molecular challenges in computational chemistry. We proposed a comprehensive categorization that encompasses two primary approaches; conventional deep learning and geometric deep learning models. This classification takes into account the distinct techniques employed by the algorithms within each approach. We present an up-to-date analysis of these algorithms, emphasizing their key features and open issues. This includes details of input descriptors, datasets used, open-source code availability, task solutions, and actual research applications, focusing on general applications rather than specific ones such as drug discovery. Furthermore, our report discusses trends and future directions in molecular algorithm design, including the input descriptors used for each deep learning model, GPU usage, training and forward processing time, model parameters, the most commonly used datasets, libraries, and optimization schemes. This information aids in identifying the most suitable algorithms for a given task. It also serves as a reference for the datasets and input data frequently used for each algorithm technique. In addition, it provides insights into the benefits and open issues of each technique, and supports the development of novel computational chemistry systems.

A deep learning method for drug-target affinity prediction based on sequence interaction information mining

Background

A critical aspect of in silico drug discovery involves the prediction of drug-target affinity (DTA). Conducting wet lab experiments to determine affinity is both expensive and time-consuming, making it necessary to find alternative approaches. In recent years, deep learning has emerged as a promising technique for DTA prediction, leveraging the substantial computational power of modern computers.

Methods

We proposed a novel sequence-based approach, named KC-DTA, for predicting drug-target affinity (DTA). In this approach, we converted the target sequence into two distinct matrices, while representing the molecule compound as a graph. The proposed method utilized k-mers analysis and Cartesian product calculation to capture the interactions and evolutionary information among various residues, enabling the creation of the two matrices for target sequence. For molecule, it was represented by constructing a molecular graph where atoms serve as nodes and chemical bonds serve as edges. Subsequently, the obtained target matrices and molecule graph were utilized as inputs for convolutional neural networks (CNNs) and graph neural networks (GNNs) to extract hidden features, which were further used for the prediction of binding affinity.

Results

In order to evaluate the effectiveness of the proposed method, we conducted several experiments and made a comprehensive comparison with the state-of-the-art approaches using multiple evaluation metrics. The results of our experiments demonstrated that the KC-DTA method achieves high performance in predicting drug-target affinity (DTA). The findings of this research underscore the significance of the KC-DTA method as a valuable tool in the field of in silico drug discovery, offering promising opportunities for accelerating the drug development process. All the data and code are available for access on https://github.com/syc2017/KCDTA.

GraphGPT: A Graph Enhanced Generative Pretrained Transformer for Conditioned Molecular Generation

Condition-based molecular generation can generate a large number of molecules with particular properties, expanding the virtual drug screening library, and accelerating the process of drug discovery. In this study, we combined a molecular graph structure and sequential representations using a generative pretrained transformer (GPT) architecture for generating molecules conditionally. The incorporation of graph structure information facilitated a better comprehension of molecular topological features, and the augmentation of a sequential contextual understanding of GPT architecture facilitated molecular generation. The experiments indicate that our model efficiently produces molecules with the desired properties, with valid and unique metrics that are close to 100%. Faced with the typical task of generating molecules based on a scaffold in drug discovery, our model is able to preserve scaffold information and generate molecules with low similarity and specified properties.

Analysis of metabolites in human gut: illuminating the design of gut-targeted drugs

Gut-targeted drugs provide a new drug modality besides that of oral, systemic molecules, that could tap into the growing knowledge of gut metabolites of bacterial or host origin and their involvement in biological processes and health through their interaction with gut targets (bacterial or host, too). Understanding the properties of gut metabolites can provide guidance for the design of gut-targeted drugs. In the present work we analyze a large set of gut metabolites, both shared with serum or present only in gut, and compare them with oral systemic drugs. We find patterns specific for these two subsets of metabolites that could be used to design drugs targeting the gut. In addition, we develop and openly share a Super Learner model to predict gut permanence, in order to aid in the design of molecules with appropriate profiles to remain in the gut, resulting in molecules with putatively reduced secondary effects and better pharmacokinetics.

In silico prediction of hERG blockers using machine learning and deep learning approaches

The human ether-à-go-go-related gene (hERG) is associated with drug cardiotoxicity. If the hERG channel is blocked, it will lead to prolonged QT interval and cause sudden death in severe cases. Therefore, it is important to evaluate the hERG-blocking property of compounds in early drug discovery. In this study, a dataset containing 4556 compounds with IC50 values determined by patch clamp techniques on mammalian lineage cells was collected, and hERG blockers and non-blockers were distinguished according to three single thresholds and two binary thresholds. Four machine learning (ML) algorithms combining four molecular fingerprints and molecular descriptors as well as graph convolutional neural networks (GCNs) were used to construct a series of binary classification models. The results showed that the best models varied for different thresholds. The ML models implemented by support vector machine and random forest performed well based on Morgan fingerprints and molecular descriptors, with AUCs ranging from 0.884 to 0.950. GCN showed superior prediction performance with AUCs above 0.952, which might be related to its direct extraction of molecular features from the original input. Meanwhile, the classification of binary threshold was better than that of single threshold, which could provide us with a more accurate prediction of hERG blockers. At last, the applicability domain for the model was defined, and seven structural alerts that might generate hERG blockage were identified by information gain and substructure frequency analysis. Our work would be beneficial for identifying hERG blockers in chemicals.

An improved multi-modal representation-learning model based on fusion networks for property prediction in drug discovery

Accurate characterization of molecular representations plays an important role in the property prediction based on deep learning (DL) for drug discovery. However, most previous researches considered only one type of molecular representations, resulting in that it difficult to capture the full molecular feature information. In this study, a novel DL framework called multi-modal molecular representation learning fusion network (MMRLFN) is developed, which could simultaneously learn and integrate drug molecular features from molecular graphs and SMILES sequences. The developed MMRLFN method is composed of three complementary deep neural networks to learn various features from different molecular representations, such as molecular topology, local chemical background information, and substructures at varying scales. Eight public datasets involving various molecular properties used in drug discovery were employed to train and evaluate the developed MMRLFN. The obtained models showed better performances than the existing models based on mono-modal molecular representations. Additionally, a thorough analysis of the noise resistance and interpretability of the MMRLFN has been carried out. The generalization ability and effectiveness of the MMRLFN has been verified by case studies as well. Overall, the MMRLFN can accurately predict molecular properties and provide potentially valuable information from large datasets, thereby maximizing the possibility of successful drug discovery.

Drug-target binding affinity prediction using message passing neural network and self supervised learning

BACKGROUND

Drug-target binding affinity (DTA) prediction is important for the rapid development of drug discovery. Compared to traditional methods, deep learning methods provide a new way for DTA prediction to achieve good performance without much knowledge of the biochemical background. However, there are still room for improvement in DTA prediction: (1) only focusing on the information of the atom leads to an incomplete representation of the molecular graph; (2) the self-supervised learning method could be introduced for protein representation.

RESULTS

In this paper, a DTA prediction model using the deep learning method is proposed, which uses an undirected-CMPNN for molecular embedding and combines CPCProt and MLM models for protein embedding. An attention mechanism is introduced to discover the important part of the protein sequence. The proposed method is evaluated on the datasets Ki and Davis, and the model outperformed other deep learning methods.

CONCLUSIONS

The proposed model improves the performance of the DTA prediction, which provides a novel strategy for deep learning-based virtual screening methods.

SubMDTA: drug target affinity prediction based on substructure extraction and multi-scale features

BACKGROUND

Drug-target affinity (DTA) prediction is a critical step in the field of drug discovery. In recent years, deep learning-based methods have emerged for DTA prediction. In order to solve the problem of fusion of substructure information of drug molecular graphs and utilize multi-scale information of protein, a self-supervised pre-training model based on substructure extraction and multi-scale features is proposed in this paper.

RESULTS

For drug molecules, the model obtains substructure information through the method of probability matrix, and the contrastive learning method is implemented on the graph-level representation and subgraph-level representation to pre-train the graph encoder for downstream tasks. For targets, a BiLSTM method that integrates multi-scale features is used to capture long-distance relationships in the amino acid sequence. The experimental results showed that our model achieved better performance for DTA prediction.

CONCLUSIONS

The proposed model improves the performance of the DTA prediction, which provides a novel strategy based on substructure extraction and multi-scale features.

Multitask learning-driven identification of novel antitrypanosomal compounds

Background: Chagas disease and human African trypanosomiasis cause substantial death and morbidity, particularly in low- and middle-income countries, making the need for novel drugs urgent. Methodology & results: Therefore, an explainable multitask pipeline to profile the activity of compounds against three trypanosomes (Trypanosoma brucei brucei, Trypanosoma brucei rhodesiense and Trypanosoma cruzi) were created. These models successfully discovered four new experimental hits (LC-3, LC-4, LC-6 and LC-15). Among them, LC-6 showed promising results, with IC50 values ranging 0.01-0.072 μM and selectivity indices >10,000. Conclusion: These results demonstrate that the multitask protocol offers predictivity and interpretability in the virtual screening of new antitrypanosomal compounds and has the potential to improve hit rates in Chagas and human African trypanosomiasis projects.

NoiseMol: A noise-robusted data augmentation via perturbing noise for molecular property prediction

Simplified Molecular-Input Line-Entry System (SMILES) is one of a widely used molecular representation methods for molecular property prediction. We conjecture that all the characters in the SMILES string of a molecule are essential for making up the molecules, but most of them make little contribution to determining a particular property of the molecule. Therefore, we verified the conjecture in the pre-experiment. Motivated by the result, we propose to inject proper noisy information into the SMILES to augment the training data by increasing the diversity of the labeled molecules. To this end, we explore injecting perturbing noise into the original labeled SMILES strings to construct augmented data for alleviating the limitation of the labeled compound data and enhancing the model to extract more useful molecular representation for molecular property prediction. Specifically, we directly adopt mask, swap, deletion, and fusion operations on SMILES strings to randomly mask, swap, and delete atoms in SMILES strings. Then, the augmented data is used by two strategies: each epoch alternately feeds the original and perturbing noisy molecules, or each batch alternately feeds the original and perturbing noisy molecules. We conduct experiments on both Transformer and BiGRU models to validate the effectiveness by adopting widely used datasets from MoleculeNet and ZINC. Experimental results demonstrate that the proposed method outperforms strong baselines on all the datasets. NoiseMol obtains the best performance on BBBP and FDA when compared with state-of-the-art methods. Besides, NoiseMol achieves the best accuracy on LogP. Therefore, injecting perturbing noise into the labeled SMILES strings is an effective and efficient method, which improves the prediction performance, generalization, and robustness of the deep learning models.

NHGNN-DTA: a node-adaptive hybrid graph neural network for interpretable drug-target binding affinity prediction

MOTIVATION

Large-scale prediction of drug-target affinity (DTA) plays an important role in drug discovery. In recent years, machine learning algorithms have made great progress in DTA prediction by utilizing sequence or structural information of both drugs and proteins. However, sequence-based algorithms ignore the structural information of molecules and proteins, while graph-based algorithms are insufficient in feature extraction and information interaction.

RESULTS

In this article, we propose NHGNN-DTA, a node-adaptive hybrid neural network for interpretable DTA prediction. It can adaptively acquire feature representations of drugs and proteins and allow information to interact at the graph level, effectively combining the advantages of both sequence-based and graph-based approaches. Experimental results have shown that NHGNN-DTA achieved new state-of-the-art performance. It achieved the mean squared error (MSE) of 0.196 on the Davis dataset (below 0.2 for the first time) and 0.124 on the KIBA dataset (3% improvement). Meanwhile, in the case of cold start scenario, NHGNN-DTA proved to be more robust and more effective with unseen inputs than baseline methods. Furthermore, the multi-head self-attention mechanism endows the model with interpretability, providing new exploratory insights for drug discovery. The case study on Omicron variants of SARS-CoV-2 illustrates the efficient utilization of drug repurposing in COVID-19.

AVAILABILITY AND IMPLEMENTATION

The source code and data are available at https://github.com/hehh77/NHGNN-DTA.

Enhancing Molecular Representations Via Graph Transformation Layers

Molecular representation learning is an essential component of many molecule-oriented tasks, such as molecular property prediction and molecule generation. In recent years, graph neural networks (GNNs) have shown great promise in this area, representing a molecule as a graph composed of nodes and edges. There are increasing studies showing that coarse-grained or multiview molecular graphs are important for molecular representation learning. Most of their models, however, are too complex and lack flexibility in learning different granular information for different tasks. Here, we proposed a flexible and simple graph transformation layer (i.e., LineEvo), a plug-and-use module for GNNs, which enables molecular representation learning from multiple perspectives. The LineEvo layer transforms fine-grained molecular graphs into coarse-grained ones based on the line graph transformation strategy. Especially, it treats the edges as nodes and generates the new connected edges, atom features, and atom positions. By stacking LineEvo layers, GNNs can learn multilevel information, from atom-level to triple-atoms level and coarser level. Experimental results show that the LineEvo layers can improve the performance of traditional GNNs on molecular property prediction benchmarks on average by 7%. Additionally, we show that the LineEvo layers can help GNNs have more expressive power than the Weisfeiler-Lehman graph isomorphism test.

Artificial intelligence for drug discovery: Resources, methods, and applications

Conventional wet laboratory testing, validations, and synthetic procedures are costly and time-consuming for drug discovery. Advancements in artificial intelligence (AI) techniques have revolutionized their applications to drug discovery. Combined with accessible data resources, AI techniques are changing the landscape of drug discovery. In the past decades, a series of AI-based models have been developed for various steps of drug discovery. These models have been used as complements of conventional experiments and have accelerated the drug discovery process. In this review, we first introduced the widely used data resources in drug discovery, such as ChEMBL and DrugBank, followed by the molecular representation schemes that convert data into computer-readable formats. Meanwhile, we summarized the algorithms used to develop AI-based models for drug discovery. Subsequently, we discussed the applications of AI techniques in pharmaceutical analysis including predicting drug toxicity, drug bioactivity, and drug physicochemical property. Furthermore, we introduced the AI-based models for de novo drug design, drug-target structure prediction, drug-target interaction, and binding affinity prediction. Moreover, we also highlighted the advanced applications of AI in drug synergism/antagonism prediction and nanomedicine design. Finally, we discussed the challenges and future perspectives on the applications of AI to drug discovery.

Molecular Property Prediction by Combining LSTM and GAT

Molecular property prediction is an important direction in computer-aided drug design. In this paper, to fully explore the information from SMILE stings and graph data of molecules, we combined the SALSTM and GAT methods in order to mine the feature information of molecules from sequences and graphs. The embedding atoms are obtained through SALSTM, firstly using SMILES strings, and they are combined with graph node features and fed into the GAT to extract the global molecular representation. At the same time, data augmentation is added to enlarge the training dataset and improve the performance of the model. Finally, to enhance the interpretability of the model, the attention layers of both models are fused together to highlight the key atoms. Comparison with other graph-based and sequence-based methods, for multiple datasets, shows that our method can achieve high prediction accuracy with good generalizability.

SuHAN: Substructural hierarchical attention network for molecular representation

Recently, molecular representation and property exploration, with the combination of neural network, play a critical role in the field of drug design and discovery for assisting in drug related research. However, previous research in molecular representation relies heavily on artificial extraction of features based on biological experiments which may result in a manually introduced noise of molecular information with high cost in time and money. In this paper, a novel method named Substructural Hierarchical Attention Network (SuHAN) is proposed to discover inherent characteristics of molecules for representation learning. Specifically, SuHAN is composed of the cascaded layer: atom-level layer and substructure-level layer. Molecule in the SMILES format is divided into several substructural fragments by predefined partition rules, and then they are fed into atom-level layer and substructure-level layer successively to obtain feature representation from different perspective: atomic view and substructural view. In this way, the prominent structural features that may be omitted in global extraction are excavated from a fine-grained viewpoint and fused to reconstruct representative pattern in an overall view. Experiments on biophysics and physiology datasets demonstrate that our model is competitive with a significant improvement of both accuracy and stability in performance. We confirmed that the substructural segments and progressive hierarchical networks lead to an effective molecular representation for downstream tasks. These results provide a novel perspective about reconstructing overall pattern through local prominent structure.

RGN: Residue-Based Graph Attention and Convolutional Network for Protein-Protein Interaction Site Prediction

The prediction of a protein-protein interaction site (PPI site) plays a very important role in the biochemical process, and lots of computational methods have been proposed in the past. However, the majority of the past methods are time consuming and lack accuracy. Hence, coming up with an effective computational method is necessary. In this article, we present a novel computational model called RGN (residue-based graph attention and convolutional network) to predict PPI sites. In our paper, the protein is treated as a graph. The amino acid can be seen as the node in the graph structure. The position-specific scoring matrix, hidden Markov model, hydrogen bond estimation algorithm, and ProtBert are applied as node features. The edges are decided by the spatial distance between the amino acids. Then, we utilize a residue-based graph convolutional network and graph attention network to further extract the deeper feature. Finally, the processed node feature is fed into the prediction layer. We show the superiority of our model by comparing it with the other four protein structure-based methods and five protein sequence-based methods. Our model obtains the best performance on all the evaluation metrics (accuracy, precision, recall, F₁ score, Matthews correlation coefficient, area under the receiver operating characteristic curve, and area under the precision recall curve). We also conduct a case study to demonstrate that extracting the protein information from the protein structure perspective is effective and points out the difficult aspect of PPI site prediction.

Computational analysis of arrhythmogenesis in KCNH2 T618I mutation-associated short QT syndrome and the pharmacological effects of quinidine and sotalol

Short QT syndrome (SQTS) is a rare but dangerous genetic disease. In this research, we conducted a comprehensive in silico investigation into the arrhythmogenesis in KCNH2 T618I-associated SQTS using a multi-scale human ventricle model. A Markov chain model of I_Kr was developed firstly to reproduce the experimental observations. It was then incorporated into cell, tissue, and organ models to explore how the mutation provided substrates for ventricular arrhythmias. Using this T618I Markov model, we explicitly revealed the subcellular level functional alterations by T618I mutation, particularly the changes of ion channel states that are difficult to demonstrate in wet experiments. The following tissue and organ models also successfully reproduced the changed dynamics of reentrant spiral waves and impaired rate adaptions in hearts of T618I mutation. In terms of pharmacotherapy, we replicated the different effects of a drug under various conditions using identical mathematical descriptions for drugs. This study not only simulated the actions of an effective drug (quinidine) at various physiological levels, but also elucidated why the I_Kr inhibitor sotalol failed in SQT1 patients through profoundly analyzing its mutation-dependent actions.