Molecular fingerprint-based artificial neural networks QSAR for ligand biological activity predictions

A New Fingerprint and Graph Hybrid Neural Network for Predicting Molecular Properties

Machine learning plays a role in accelerating drug discovery, and the design of effective machine learning models is crucial for accurately predicting molecular properties. Characterizing molecules typically involves the use of molecular fingerprints and molecular graphs. These are input into a multilayer perceptron (MLP) and variants of graph neural networks, such as graph attention networks (GATs). Due to the diverse types and large dimension of fingerprints, models may contain many features that are relatively irrelevant or redundant; meanwhile, although the GAT excels in handling heterogeneous graph tasks, it lacks the ability to extract collaborative information from neighboring nodes, which is crucial in scenarios where it cannot capture the joint influence of adjacent groups on atoms. To overcome these challenges, we introduce a hybrid model, combining improved GAT and MLP. In GAT, the recurrent neural network is employed to capture collaborative information. To address the dimensionality issue, we propose a feature selection algorithm, which is based on the principle of maximizing relevance while minimizing redundancy. Through experiments on 13 public data sets and 14 breast cell lines, our model demonstrates superior performance compared to state-of-the-art deep learning and traditional machine learning algorithms. Additionally, a series of ablation experiments were conducted to demonstrate the advantages of our improved version, as well as its antinoise capability and interpretability. These results indicate that our model holds promising prospects for practical applications.

Efficient clustering of large molecular libraries

The widespread use of Machine Learning (ML) techniques in chemical applications has come with the pressing need to analyze extremely large molecular libraries. In particular, clustering remains one of the most common tools to dissect the chemical space. Unfortunately, most current approaches present unfavorable time and memory scaling, which makes them unsuitable to handle million- and billion-sized sets. Here, we propose to bypass these problems with a time- and memory-efficient clustering algorithm, BitBIRCH. This method uses a tree structure similar to the one found in the Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) algorithm to ensure O N time scaling. BitBIRCH leverages the instant similarity (iSIM) formalism to process binary fingerprints, allowing the use of Tanimoto similarity, and reducing memory requirements. Our tests show that BitBIRCH is already > 1,000 times faster than standard implementations of the Taylor-Butina clustering for libraries with 1,500,000 molecules. BitBIRCH increases efficiency without compromising the quality of the resulting clusters. We explore strategies to handle large sets, which we applied in the clustering of one billion molecules under 5 hours using a parallel/iterative BitBIRCH approximation.

A novel interpretable machine learning model approach for the prediction of TiO2 photocatalytic degradation of air contaminants

Air contaminants lead to various environmental and health issues. Titanium dioxide (TiO2) features the benefits of autogenous photocatalytic degradation of air contaminants. To evaluate its performance, laboratory experiments are commonly used to determine the kinetics of the photocatalytic-degradation rate, which is labor intensive, time-consuming, and costly. In this study, Machine Learning (ML) models were developed to predict the photo-degradation rate constants of air-borne organic contaminants with TiO2 nanoparticles and ultraviolet irradiation. The hyperparameters of the ML models were optimized, which included Artificial Neural Network (ANN) with Bayesian optimization, gradient booster regressor (GBR) with Bayesian optimization, Extreme Gradient Boosting (XGBoost) with optimization using Hyperopt, and Catboost combined with Adaboost. The organic contaminant was encoded through Molecular fingerprints (MF). Imputation method was applied to deal with the missing data. A generative ML model Vanilla Gan was utilized to create synthetic data to further augment the size of available dataset and the SHapley Additive exPlanations (SHAP) was employed for ML model interpretability. The results indicated that data imputation allowed for the full utilization of the limited dataset, leading to good machine learning prediction performance and preventing common overfitting problems with small-sized data. Additionally, augmenting experimental data with synthetic data significantly improved prediction accuracy and considerably reduced overfitting issues. The results ranked the feature importance and assessed the impacts of different experimental variables on the rate of photo-degradation, which were consistent with physico-chemical laws.

Chemodivergence in Fluorine Source-Controlled Cascade Reaction of Aryne Precursors to Synthesize Pyrrolo[3,4-b]indoles and 3-Arylated Maleimides

Reactions allowing chemodivergence prove to be attractive strategies in synthetic organic chemistry. We herein described a highly practical, transition-metal-free, highly regioselective and chemodivergent cascade reaction controlled by fluorine sources, which involved a [3 + 2] cycloaddition or C-arylation process between aryne precursors and 3-aminomaleimides. These two pathways led to a wide scope of structurally diverse pyrrolo[3,4-b]indoles (19 examples) and 3-arylated maleimides (25 examples) in good-to-excellent yields. Furthermore, the reaction could be scaled up, and several synthetic transformations were accomplished for the preparation of functionalized molecules and might provide new opportunities for the discovery of N-heterocyclic drugs.

First report on chemometrics-driven multilayered lead prioritization in addressing oxysterol-mediated overexpression of G protein-coupled receptor 183

Contemporary research has convincingly demonstrated that upregulation of G protein-coupled receptor 183 (GPR183), orchestrated by its endogenous agonist, 7α,25-dihydroxyxcholesterol (7α,25-OHC), leads to the development of cancer, diabetes, multiple sclerosis, infectious, and inflammatory diseases. A recent study unveiled the cryo-EM structure of 7α,25-OHC bound GPR183 complex, presenting an untapped opportunity for computational exploration of potential GPR183 inhibitors, which served as our inspiration for the current work. A predictive and validated two-dimensional QSAR model using genetic algorithm (GA) and multiple linear regression (MLR) on experimental GPR183 inhibition data was developed. QSAR study highlighted that structural features like dissimilar electronegative atoms, quaternary carbon atoms, and CH2RX fragment (X: heteroatoms) influence positively, while the existence of oxygen atoms with a topological separation of 3, negatively affects GPR183 inhibitory activity. Post assessment of true external set prediction capability, the MLR model was deployed to screen 12,449 DrugBank compounds, followed by a screening pipeline involving molecular docking, druglikeness, ADMET, protein-ligand stability assessment using deep learning algorithm, molecular dynamics, and molecular mechanics. The current findings strongly evidenced DB05790 as a potential lead for prospective interference of oxysterol-mediated GPR183 overexpression, warranting further in vitro and in vivo validation.

Advancing chronic toxicity risk assessment in freshwater ecology by molecular characterization-based machine learning

The continuously increased production of various chemicals and their release into environments have raised potential negative effects on ecological health. However, traditional labor-intensive assessment methods cannot effectively and rapidly evaluate these hazards, especially for chronic risk. In this study, machine learning (ML) was employed to construct quantitative structure-activity relationship (QSAR) models, enabling the prediction of chronic toxicity to aquatic organisms by leveraging the molecular characteristics of pollutants, namely, the molecular descriptors, fingerprints, and graphs. The limited dataset size hindered the notable advantages of the graph attention network (GAT) model for the molecular graphs. Considering computational efficiency and performance (R2 = 0.78; RMSE = 0.77), XGBoost (XGB) was used for reliable QSAR-ML models predicting chronic toxicity using small- or medium-sized tabular data and the molecular descriptors. Further kernel density estimation analysis confirmed the high accuracy of the model for pollutant concentrations ranging from 10-3 to 102 mg/L, effectively aligning with most environmental scenarios. Model interpretation showed SlogP and exposure duration as the primary influential factors. SlogP, representing the distribution coefficient of a molecule between lipophilic and hydrophilic environments, had a negative effect on the toxicity outcomes. Additionally, the exposure duration played a crucial role in determining the chronic toxicity. Finally, the chronic toxicity data of bisphenol A validated the robustness and reliability of the model established in this research. Our study provided a robust and feasible methodology for chronic ecological risk evaluation of various types of pollutants and could facilitate and increase the use of ML applications in environmental fields.

Attention Mechanism-Based Graph Neural Network Model for Effective Activity Prediction of SARS-CoV-2 Main Protease Inhibitors: Application to Drug Repurposing as Potential COVID-19 Therapy

Compared to de novo drug discovery, drug repurposing provides a time-efficient way to treat coronavirus disease 19 (COVID-19) that is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). SARS-CoV-2 main protease (Mpro) has been proved to be an attractive drug target due to its pivotal involvement in viral replication and transcription. Here, we present a graph neural network-based deep-learning (DL) strategy to prioritize the existing drugs for their potential therapeutic effects against SARS-CoV-2 Mpro. Mpro inhibitors were represented as molecular graphs ready for graph attention network (GAT) and graph isomorphism network (GIN) modeling for predicting the inhibitory activities. The result shows that the GAT model outperforms the GIN and other competitive models and yields satisfactory predictions for unseen Mpro inhibitors, confirming its robustness and generalization. The attention mechanism of GAT enables to capture the dominant substructures and thus to realize the interpretability of the model. Finally, we applied the optimal GAT model in conjunction with molecular docking simulations to screen the Drug Repurposing Hub (DRH) database. As a result, 18 drug hits with best consensus prediction scores and binding affinity values were identified as the potential therapeutics against COVID-19. Both the extensive literature searching and evaluations on adsorption, distribution, metabolism, excretion, and toxicity (ADMET) illustrate the premium drug-likeness and pharmacokinetic properties of the drug candidates. Overall, our work not only provides an effective GAT-based DL prediction tool for inhibitory activity of SARS-CoV-2 Mpro inhibitors but also provides theoretical guidelines for drug discovery in the COVID-19 treatment.

Count-Based Morgan Fingerprint: A More Efficient and Interpretable Molecular Representation in Developing Machine Learning-Based Predictive Regression Models for Water Contaminants' Activities and Properties

In this study, we introduce the count-based Morgan fingerprint (C-MF) to represent chemical structures of contaminants and develop machine learning (ML)-based predictive models for their activities and properties. Compared with the binary Morgan fingerprint (B-MF), C-MF not only qualifies the presence or absence of an atom group but also quantifies its counts in a molecule. We employ six different ML algorithms (ridge regression, SVM, KNN, RF, XGBoost, and CatBoost) to develop models on 10 contaminant-related data sets based on C-MF and B-MF to compare them in terms of the model's predictive performance, interpretation, and applicability domain (AD). Our results show that C-MF outperforms B-MF in nine of 10 data sets in terms of model predictive performance. The advantage of C-MF over B-MF is dependent on the ML algorithm, and the performance enhancements are proportional to the difference in the chemical diversity of data sets calculated by B-MF and C-MF. Model interpretation results show that the C-MF-based model can elucidate the effect of atom group counts on the target and have a wider range of SHAP values. AD analysis shows that C-MF-based models have an AD similar to that of B-MF-based ones. Finally, we developed a "ContaminaNET" platform to deploy these C-MF-based models for free use.

Augmenting Polymer Datasets by Iterative Rearrangement

One of the biggest obstacles to successful polymer property prediction is an effective representation that accurately captures the sequence of repeat units in a polymer. Motivated by the success of data augmentation in computer vision and natural language processing, we explore augmenting polymer data by iteratively rearranging the molecular representation while preserving the correct connectivity, revealing additional substructural information that is not present in a single representation. We evaluate the effects of this technique on the performance of machine learning models trained on three polymer datasets and compare them to common molecular representations. Data augmentation does not yield significant improvements in machine learning property prediction performance compared to equivalent (non-augmented) representations. In datasets where the target property is primarily influenced by the polymer sequence rather than experimental parameters, this data augmentation technique provides molecular embedding with more information to improve property prediction accuracy.

A consensual machine-learning-assisted QSAR model for effective bioactivity prediction of xanthine oxidase inhibitors using molecular fingerprints

Xanthine oxidase inhibitors (XOIs) have been widely studied due to the promising potential as safe and effective therapeutics in hyperuricemia and gout. Currently, available XOI molecules have been developed from different experiments but they are with the wide structure diversity and significant varying bioactivities. So it is of great practical significance to present a consensual QSAR model for effective bioactivity prediction of XOIs based on a systematic compiling of these XOIs across different experiments. In this work, 249 XOIs belonging to 16 scaffolds were collected and were integrated into a consensual dataset by introducing the concept of IC₅₀ values relative to allopurinol (RIC₅₀). Here, extended connectivity fingerprints (ECFPs) were employed to represent XOI molecules. By performing effective feature selection by machine-learning method, 54 crucial fingerprints were indicated to be valuable for predicting the inhibitory potency (IP) of XOIs. The optimal predictor yields the promising performance by different cross-validation tests. Besides, an external validation of 43 XOIs and a case study on febuxostat also provide satisfactory results, indicating the powerful generalization of our predictor. Here, the predictor was interpreted by shapely additive explanation (SHAP) method which revealed several important substructures by mapping the featured fingerprints to molecular structures. Then, 15 new molecules were designed and predicted by our predictor to show superior IP than febuxostat. Finally, molecular docking simulation was performed to gain a deep insight into molecular binding mode with xanthine oxidase (XO) enzyme, showing that molecules with selenazole moiety, cyano group and isopropyl group tended to yield higher IP. The absorption, distribution, metabolism, excretion and toxicity (ADMET) prediction results further enhanced the potential of these novel XOIs as drug candidates. Overall, this work presents a QSAR model for accurate prediction of IP of XOIs, and is expected to provide new insights for further structure-guided design of novel XOIs.

Novel Thiosemicarbazone Quantum Dots in the Treatment of Alzheimer's Disease Combining In Silico Models Using Fingerprints and Physicochemical Descriptors

Searching for thiosemicarbazone derivatives with the potential to inhibit acetylcholinesterase for the treatment of Alzheimer's disease (AD) is an important current goal. The QSAR_KPLS, QSAR_ANN, and QSAR_SVR models were constructed using binary fingerprints and physicochemical (PC) descriptors of 129 thiosemicarbazone compounds screened from a database of 3791 derivatives. The R ² and Q ² values for the QSAR_KPLS, QSAR_ANN, and QSAR_SVR models are greater than 0.925 and 0.713 using dendritic fingerprint (DF) and PC descriptors, respectively. The in vitro pIC₅₀ activities of four new design-oriented compounds N1, N2, N3, and N4, from the QSAR_KPLS model using DFs, are consistent with the experimental results and those from the QSAR_ANN and QSAR_SVR models. The designed compounds N1, N2, N3, and N4 do not violate Lipinski-5 and Veber rules using the ADME and BoiLED-Egg methods. The binding energy, kcal mol^-1, of the novel compounds to the 1ACJ-PDB protein receptor of the AChE enzyme was also obtained by molecular docking and dynamics simulations consistent with those predicted from the QSAR_ANN and QSAR_SVR models. New compounds N1, N2, N3, and N4 were synthesized, and the experimental in vitro pIC₅₀ activity was determined in agreement with those obtained from in silico models. The newly synthesized thiosemicarbazones N1, N2, N3, and N4 can inhibit 1ACJ-PDB, which is predicted to be able to cross the barrier. The DFT B3LYP/def-SV(P)-ECP quantization calculation method was used to calculate E _HOMO and E _LUMO to account for the activities of compounds N1, N2, N3, and N4. The quantum calculation results explained are consistent with those obtained in in silico models. The successful results here may contribute to the search for new drugs for the treatment of AD.

Biological Activity Predictions of Ligands Based on Hybrid Molecular Fingerprinting and Ensemble Learning

The biological activity predictions of ligands are an important research direction, which can improve the efficiency and success probability of drug screening. However, the traditional prediction method has the disadvantages of complex modeling and low screening efficiency. Machine learning is considered an important research direction to solve these traditional method problems in the near future. This paper proposes a machine learning model with high predictive accuracy and stable prediction ability, namely, the back propagation neural network cross-support vector regression model (BPCSVR). By comparing multiple molecular descriptors, MACCS fingerprint and ECFP6 fingerprint were selected as inputs, and the stable prediction ability of the model was improved by integrating multiple models and correcting similar samples. We used leave-one-out cross-validation on 3038 samples from six data sets. The coefficient of determination, root mean square error, and absolute error were used as the evaluation parameters. After comparing the multiclass models, the results show that the BPCSVR model has stable prediction ability in different data sets, and the prediction accuracy is higher than other comparison models.

Inverse QSAR: Reversing Descriptor-Driven Prediction Pipeline Using Attention-Based Conditional Variational Autoencoder

In order to better foramize it, the notorious inverse-QSAR problem (finding structures of given QSAR-predicted properties) is considered in this paper as a two-step process including (i) finding "seed" descriptor vectors corresponding to user-constrained QSAR model output values and (ii) identifying the chemical structures best matching the "seed" vectors. The main development effort here was focused on the latter stage, proposing a new attention-based conditional variational autoencoder neural-network architecture based on recent developments in attention-based methods. The obtained results show that this workflow was capable of generating compounds predicted to display desired activity while being completely novel compared to the training database (ChEMBL). Moreover, the generated compounds show acceptable druglikeness and synthetic accessibility. Both pharmacophore and docking studies were carried out as "orthogonal" in silico validation methods, proving that some of de novo structures are, beyond being predicted active by 2D-QSAR models, clearly able to match binding 3D pharmacophores and bind the protein pocket.

Quantitative Structure-Toxicity Relationship in Bioactive Molecules from a Conceptual DFT Perspective

The preclinical drug discovery stage often requires a large amount of costly and time-consuming experiments using huge sets of chemical compounds. In the last few decades, this process has undergone significant improvements by the introduction of quantitative structure-activity relationship (QSAR) modelling that uses a certain percentage of experimental data to predict the biological activity/property of compounds with similar structural skeleton and/or containing a particular functional group(s). The use of machine learning tools along with it has made life even easier for pharmaceutical researchers. Here, we discuss the toxicity of certain sets of bioactive compounds towards Pimephales promelas and Tetrahymena pyriformis in terms of the global conceptual density functional theory (CDFT)-based descriptor, electrophilicity index (ω). We have compared the results with those obtained by using the commonly used hydrophobicity parameter, logP (where P is the n-octanol/water partition coefficient), considering the greater ease of computing the ω descriptor. The Human African trypanosomiasis (HAT) curing activity of 32 pyridyl benzamide derivatives is also studied against Tryphanosoma brucei. In this review article, we summarize these multiple linear regression (MLR)-based QSAR studies in terms of electrophilicity (ω, ω2) and hydrophobicity (logP, (logP)2) parameters.

Concepts and applications of chemical fingerprint for hit and lead screening

Molecular fingerprints are used to represent chemical (structural, physicochemical, etc.) properties of large-scale chemical sets in a low computational cost way. They have a prominent role in transforming chemical data sets into consistent input formats (bit strings or numeric values) suitable for in silico approaches. In this review, we summarize and classify common and state-of-the-art fingerprints into eight different types (dictionary based, circular, topological, pharmacophore, protein-ligand interaction, shape based, reinforced, and multi). We also highlight applications of fingerprints in early drug research and development (R&D). Thus, this review provides a guide for the selection of appropriate fingerprints of compounds (or ligand-protein complexes) for use in drug R&D.

Predicting mosquito repellents for clothing application from molecular fingerprint-based artificial neural network SAR models

Spraying repellents on clothing limits toxicity and allergy problems that can occur when the repellents are directly applied to skin. This also allows the use of higher doses to ensure longer lasting effects. As the number of repellents available on the market is limited, it is necessary to propose new ones, especially by using in silico methods that reduce costs and time. In this context SAR models were built from a dataset of 2027 chemicals for which repellent activity on clothing was measured against Aedes aegypti. The interest of using either the ECFP or MACCS fingerprints as input neurons of a three-layer perceptron was evaluated. Transformation of MACCS bit strings into disjunctive tables led to interesting results. Models obtained with both types of fingerprints were compared to a model including physicochemical and topological descriptors.

Predicting protein network topology clusters from chemical structure using deep learning

Comparing chemical structures to infer protein targets and functions is a common approach, but basing comparisons on chemical similarity alone can be misleading. Here we present a methodology for predicting target protein clusters using deep neural networks. The model is trained on clusters of compounds based on similarities calculated from combined compound-protein and protein-protein interaction data using a network topology approach. We compare several deep learning architectures including both convolutional and recurrent neural networks. The best performing method, the recurrent neural network architecture MolPMoFiT, achieved an F1 score approaching 0.9 on a held-out test set of 8907 compounds. In addition, in-depth analysis on a set of eleven well-studied chemical compounds with known functions showed that predictions were justifiable for all but one of the chemicals. Four of the compounds, similar in their molecular structure but with dissimilarities in their function, revealed advantages of our method compared to using chemical similarity.

Artificial intelligence and machine-learning approaches in structure and ligand-based discovery of drugs affecting central nervous system

CNS disorders are indications with a very high unmet medical needs, relatively smaller number of available drugs, and a subpar satisfaction level among patients and caregiver. Discovery of CNS drugs is extremely expensive affair with its own unique challenges leading to extremely high attrition rates and low efficiency. With explosion of data in information age, there is hardly any aspect of life that has not been touched by data driven technologies such as artificial intelligence (AI) and machine learning (ML). Drug discovery is no exception, emergence of big data via genomic, proteomic, biological, and chemical technologies has driven pharmaceutical giants to collaborate with AI oriented companies to revolutionise drug discovery, with the goal of increasing the efficiency of the process. In recent years many examples of innovative applications of AI and ML techniques in CNS drug discovery has been reported. Research on therapeutics for diseases such as schizophrenia, Alzheimer's and Parkinsonism has been provided with a new direction and thrust from these developments. AI and ML has been applied to both ligand-based and structure-based drug discovery and design of CNS therapeutics. In this review, we have summarised the general aspects of AI and ML from the perspective of drug discovery followed by a comprehensive coverage of the recent developments in the applications of AI/ML techniques in CNS drug discovery.

Antibacterial Activity Prediction of Plant Secondary Metabolites Based on a Combined Approach of Graph Clustering and Deep Neural Network

The plants produce numerous types of secondary metabolites which have pharmacological importance in drug development for different diseases. Computational methods widely use the fingerprints of the metabolites to understand different properties and similarities among metabolites and for the prediction of chemical reactions etc. In this work, we developed three different deep neural network models (DNN) to predict the antibacterial property of plant metabolites. We developed the first DNN model using the fingerprint set of metabolites as features. In the second DNN model, we searched the similarities among fingerprints using correlation and used one representative feature from each group of highly correlated fingerprints. In the third model, the fingerprints of metabolites were used to find structurally similar chemical compound clusters. Form each cluster a representative metabolite is selected and made part of the training dataset. The second model reduced the number of features where the third model achieved better classification results for test data. In both cases, we applied the simple graph clustering method to cluster the corresponding network. The correlation-based DNN model reduced some features while retaining an almost similar performance compared to the first DNN model. The third model improves classification results for test data by capturing wider variance within training data using graph clustering method. This third model is somewhat novel approach and can be applied to build DNN models for other purposes.

In silico prediction of mosquito repellents for clothing application

Use of protective clothing is a simple and efficient way to reduce the contacts with mosquitoes and consequently the probability of transmission of diseases spread by them. This mechanical barrier can be enhanced by the application of repellents. Unfortunately the number of available repellents is limited. As a result, there is a crucial need to find new active and safer molecules repelling mosquitoes. In this context, a structure-activity relationship (SAR) model was proposed for the design of repellents active on clothing. It was computed from a dataset of 2027 chemicals for which repellent activity on clothing was measured against Aedes aegypti. Molecules were described by means of 20 molecular descriptors encoding physicochemical properties, topological information and structural features. A three-layer perceptron was used as statistical tool. An accuracy of 87% was obtained for both the training and test sets. Most of the wrong predictions can be explained. Avenues for increasing the performances of the model have been proposed.

Virtual screening of dipeptidyl peptidase-4 inhibitors using quantitative structure-activity relationship-based artificial intelligence and molecular docking of hit compounds

Dipeptidyl peptidase-4 (DPP-4) inhibitors are becoming an essential drug in the treatment of type 2 diabetes mellitus; however, some classes of these drugs exert side effects, including joint pain and pancreatitis. Studies suggest that these side effects might be related to secondary inhibition of DPP-8 and DPP-9. In this study, we identified DPP-4-inhibitor hit compounds selective against DPP-8 and DPP-9. We built a virtual screening workflow using a quantitative structure-activity relationship (QSAR) strategy based on artificial intelligence to allow faster screening of millions of molecules for the DPP-4 target relative to other screening methods. Five regression machine learning algorithms and four classification machine learning algorithms were applied to build virtual screening workflows, with the QSAR model applied using support vector regression (R²_pred 0.78) and the classification QSAR model using the random forest algorithm with 92.2% accuracy. Virtual screening results of > 10 million molecules obtained 2 716 hits compounds with a pIC₅₀ value of > 7.5. Additionally, molecular docking results of several potential hit compounds for DPP-4, DPP-8, and DPP-9 identified CH0002 as showing high inhibitory potential against DPP-4 and low inhibitory potential for DPP-8 and DPP-9 enzymes. These results demonstrated the effectiveness of this technique for identifying DPP-4-inhibitor hit compounds selective for DPP-4 and against DPP-8 and DPP-9 and suggest its potential efficacy for applications to discover hit compounds of other targets.

A general QSPR protocol for the prediction of atomic/inter-atomic properties: a fragment based graph convolutional neural network (F-GCN)

In this study, a general quantitative structure-property relationship (QSPR) protocol, fragment based graph convolutional neural network (F-GCN), was developed for the prediction of atomic/inter-atomic properties. We applied this novel artificial intelligence (AI) tool in predictions of NMR chemical shifts and bond dissociation energies (BDEs). The obtained results were comparable to experimental measurements, while the computational cost was substantially reduced, with respect to pure density functional theory (DFT) calculations. The two important features of F-GCN can be summarised as: first, it could utilise different levels of molecular fragments for atomic/inter-atomic information extraction; second, the designed architecture is also open to include additional descriptors for a more accurate solution of the local environment at atomic level, making itself more efficient for structural solutions. And during our test, the averaged prediction error of 1H NMR chemical shifts is as small as 0.32 ppm, and the error of C-H BDE estimation is 2.7 kcal mol-1. Moreover, we further demonstrated the applicability of this developed F-GCN model via several challenging structural assignments. The success of the F-GCN in atomic and inter-atomic predictions also indicates an essential improvement of computational chemistry with the assistance of AI tools.

Artificial intelligence and machine learning-aided drug discovery in central nervous system diseases: State-of-the-arts and future directions

Neurological disorders significantly outnumber diseases in other therapeutic areas. However, developing drugs for central nervous system (CNS) disorders remains the most challenging area in drug discovery, accompanied with the long timelines and high attrition rates. With the rapid growth of biomedical data enabled by advanced experimental technologies, artificial intelligence (AI) and machine learning (ML) have emerged as an indispensable tool to draw meaningful insights and improve decision making in drug discovery. Thanks to the advancements in AI and ML algorithms, now the AI/ML-driven solutions have an unprecedented potential to accelerate the process of CNS drug discovery with better success rate. In this review, we comprehensively summarize AI/ML-powered pharmaceutical discovery efforts and their implementations in the CNS area. After introducing the AI/ML models as well as the conceptualization and data preparation, we outline the applications of AI/ML technologies to several key procedures in drug discovery, including target identification, compound screening, hit/lead generation and optimization, drug response and synergy prediction, de novo drug design, and drug repurposing. We review the current state-of-the-art of AI/ML-guided CNS drug discovery, focusing on blood-brain barrier permeability prediction and implementation into therapeutic discovery for neurological diseases. Finally, we discuss the major challenges and limitations of current approaches and possible future directions that may provide resolutions to these difficulties.

Drug design of new 5-HT6R antagonists aided by artificial neural networks

Alzheimer's Disease (AD) is the most frequent illness and cause of death amongst the age related-neurodegenerative disorders. The Alzheimer's Disease International (ADI) reported in 2019 that over 50 million people were living with dementia in the world and this number could potentially be around 152 million by 2050.5-hydroxtryptamine subtype 6 receptor (5-HT₆R) has been identified as a potential anti-amnesic drug target and therefore, the administration of 5-HT₆R antagonists can likely mitigate the memory loss and intellectual deterioration associated with AD. Herein, computational tools were applied to design new 5-HT₆ antagonists and their biological activity values were predicted by our QSAR model obtained from Artificial Neural Networks (ANN). The proposed compounds here from the QSAR-ANN model presented significant biological activity values and some of them have achieved pK_i above 9.00. Furthermore, our results suggest that the presence of halogen atoms (especially bromine) linked to the aromatic ring at para-position (HYD) contribute considerably to the increase of the biological activity values while bulky groups in the PI position do not culminate with the increase antagonist activity of compounds here analyzed. Finally, the ADME/Tox profile as well as the synthetic accessibility of new proposed compounds qualify them to go on further with experimental procedures and thenceforward their antagonist effects can be confirmed.

Improvement of Prediction Performance With Conjoint Molecular Fingerprint in Deep Learning

The accurate predicting of physical properties and bioactivity of drug molecules in deep learning depends on how molecules are represented. Many types of molecular descriptors have been developed for quantitative structure-activity/property relationships quantitative structure-activity relationships (QSPR). However, each molecular descriptor is optimized for a specific application with encoding preference. Considering that standalone featurization methods may only cover parts of information of the chemical molecules, we proposed to build the conjoint fingerprint by combining two supplementary fingerprints. The impact of conjoint fingerprint and each standalone fingerprint on predicting performance was systematically evaluated in predicting the logarithm of the partition coefficient (logP) and binding affinity of protein-ligand by using machine learning/deep learning (ML/DL) methods, including random forest (RF), support vector regression (SVR), extreme gradient boosting (XGBoost), long short-term memory network (LSTM), and deep neural network (DNN). The results demonstrated that the conjoint fingerprint yielded improved predictive performance, even outperforming the consensus model using two standalone fingerprints among four out of five examined methods. Given that the conjoint fingerprint scheme shows easy extensibility and high applicability, we expect that the proposed conjoint scheme would create new opportunities for continuously improving predictive performance of deep learning by harnessing the complementarity of various types of fingerprints.

A generalized predictive model for TiO2-Catalyzed photo-degradation rate constants of water contaminants through artificial neural network

Titanium dioxide (TiO₂) is a well-known photocatalyst in the applications of water contaminant treatment. Traditionally, the kinetics of photo-degradation rates are obtained from experiments, which consumes enormous labor and experimental investments. Here, a generalized predictive model was developed for prediction of the photo-degradation rate constants of organic contaminants in the presence of TiO₂ nanoparticles and ultraviolet irradiation in aqueous solution. This model combines an artificial neural network (ANN) with a variety of factors that affect the photo-degradation performance, i.e., ultraviolet intensity, TiO₂ dosage, organic contaminant type and initial concentration in water, and initial pH of the solution. The molecular fingerprints (MF) were used to interpret the organic contaminants as binary vectors, a format that is machine-readable in computational linguistics. A dataset of 446 data points for training and testing was collected from the literature. This predictive model shows a good accuracy with a root mean square error (RMSE) of 0.173.

Practical Applications of Deep Learning To Impute Heterogeneous Drug Discovery Data

Contemporary deep learning approaches still struggle to bring a useful improvement in the field of drug discovery because of the challenges of sparse, noisy, and heterogeneous data that are typically encountered in this context. We use a state-of-the-art deep learning method, Alchemite, to impute data from drug discovery projects, including multitarget biochemical activities, phenotypic activities in cell-based assays, and a variety of absorption, distribution, metabolism, and excretion (ADME) endpoints. The resulting model gives excellent predictions for activity and ADME endpoints, offering an average increase in R² of 0.22 versus quantitative structure-activity relationship methods. The model accuracy is robust to combining data across uncorrelated endpoints and projects with different chemical spaces, enabling a single model to be trained for all compounds and endpoints. We demonstrate improvements in accuracy on the latest chemistry and data when updating models with new data as an ongoing medicinal chemistry project progresses.

Different molecular enumeration influences in deep learning: an example using aqueous solubility

Aqueous solubility is the key property driving many chemical and biological phenomena and impacts experimental and computational attempts to assess those phenomena. Accurate prediction of solubility is essential and challenging, even with modern computational algorithms. Fingerprint-based, feature-based and molecular graph-based representations have all been used with different deep learning methods for aqueous solubility prediction. It has been clearly demonstrated that different molecular representations impact the model prediction and explainability. In this work, we reviewed different representations and also focused on using graph and line notations for modeling. In general, one canonical chemical structure is used to represent one molecule when computing its properties. We carefully examined the commonly used simplified molecular-input line-entry specification (SMILES) notation representing a single molecule and proposed to use the full enumerations in SMILES to achieve better accuracy. A convolutional neural network (CNN) was used. The full enumeration of SMILES can improve the presentation of a molecule and describe the molecule with all possible angles. This CNN model can be very robust when dealing with large datasets since no additional explicit chemistry knowledge is necessary to predict the solubility. Also, traditionally it is hard to use a neural network to explain the contribution of chemical substructures to a single property. We demonstrated the use of attention in the decoding network to detect the part of a molecule that is relevant to solubility, which can be used to explain the contribution from the CNN.

The Study on the hERG Blocker Prediction Using Chemical Fingerprint Analysis

Human ether-a-go-go-related gene (hERG) potassium channel blockage by small molecules may cause severe cardiac side effects. Thus, it is crucial to screen compounds for activity on the hERG channels early in the drug discovery process. In this study, we collected 5299 hERG inhibitors with diverse chemical structures from a number of sources. Based on this dataset, we evaluated different machine learning (ML) and deep learning (DL) algorithms using various integer and binary type fingerprints. A training set of 3991 compounds was used to develop quantitative structure-activity relationship (QSAR) models. The performance of the developed models was evaluated using a test set of 998 compounds. Models were further validated using external set 1 (263 compounds) and external set 2 (47 compounds). Overall, models with integer type fingerprints showed better performance than models with no fingerprints, converted binary type fingerprints or original binary type fingerprints. Comparison of ML and DL algorithms revealed that integer type fingerprints are suitable for ML, whereas binary type fingerprints are suitable for DL. The outcomes of this study indicate that the rational selection of fingerprints is important for hERG blocker prediction.

Pushing property limits in materials discovery via boundless objective-free exploration

Our developed algorithm, BLOX (BoundLess Objective-free eXploration), successfully found “out-of-trend” molecules potentially useful for photofunctional materials from a drug database.

Materials chemists develop chemical compounds to meet often conflicting demands of industrial applications. This process may not be properly modeled by black-box optimization because the target property is not well defined in some cases. Herein, we propose a new algorithm for automated materials discovery called BoundLess Objective-free eXploration (BLOX) that uses a novel criterion based on kernel-based Stein discrepancy in the property space. Unlike other objective-free exploration methods, a boundary for the materials properties is not needed; hence, BLOX is suitable for open-ended scientific endeavors. We demonstrate the effectiveness of BLOX by finding light-absorbing molecules from a drug database. Our goal is to minimize the number of density functional theory calculations required to discover out-of-trend compounds in the intensity–wavelength property space. Using absorption spectroscopy, we experimentally verified that eight compounds identified as outstanding exhibit the expected optical properties. Our results show that BLOX is useful for chemical repurposing, and we expect this search method to have numerous applications in various scientific disciplines.

Open-source chemogenomic data-driven algorithms for predicting drug-target interactions

While novel technologies such as high-throughput screening have advanced together with significant investment by pharmaceutical companies during the past decades, the success rate for drug development has not yet been improved prompting researchers looking for new strategies of drug discovery. Drug repositioning is a potential approach to solve this dilemma. However, experimental identification and validation of potential drug targets encoded by the human genome is both costly and time-consuming. Therefore, effective computational approaches have been proposed to facilitate drug repositioning, which have proved to be successful in drug discovery. Doubtlessly, the availability of open-accessible data from basic chemical biology research and the success of human genome sequencing are crucial to develop effective in silico drug repositioning methods allowing the identification of potential targets for existing drugs. In this work, we review several chemogenomic data-driven computational algorithms with source codes publicly accessible for predicting drug-target interactions (DTIs). We organize these algorithms by model properties and model evolutionary relationships. We re-implemented five representative algorithms in R programming language, and compared these algorithms by means of mean percentile ranking, a new recall-based evaluation metric in the DTI prediction research field. We anticipate that this review will be objective and helpful to researchers who would like to further improve existing algorithms or need to choose appropriate algorithms to infer potential DTIs in the projects. The source codes for DTI predictions are available at: https://github.com/minghao2016/chemogenomicAlg4DTIpred.

A deep neural network combined with molecular fingerprints (DNN-MF) to develop predictive models for hydroxyl radical rate constants of water contaminants

This work combined a Deep Neural Network (DNN) with molecular fingerprints (MF) to develop models to predict the OH radical rate constants of 593 organic contaminants. Molecular descriptors, most often used in establishing quantitative structural-activity relationships (QSARs), were not used here because of their complicated generation processes that rely on advanced physicochemical and computational knowledge. Instead, we only fed the most basic information of the contaminant structures, i.e., MF encoding the types of atoms and how they are connected, to DNN and DNN then developed predictive models automatically. Here, a dataset containing 457 contaminants and their OH rate constants was first used to develop predictive models by DNN-MF. The hence developed models showed comparable accuracy to the traditional QSARs. The root mean square error (RMSE) values of the test sets were 0.358-0.384. The length of 2048 bits for the MF and 3 hidden layers (each with 1024 neurons) were found to be the optimal parameters for DNN. The model containing additional 89 micorpollutants in the training set was then successfully applied to predict the OH rate constants of 17 organophosphorus flame retardants and 29 additional micropollutants, with comparable accuracy to the reported molecular descriptors-based QSARs.

Boosting Tree-Assisted Multitask Deep Learning for Small Scientific Datasets

Machine learning approaches have had tremendous success in various disciplines. However, such success highly depends on the size and quality of datasets. Scientific datasets are often small and difficult to collect. Currently, improving machine learning performance for small scientific datasets remains a major challenge in many academic fields, such as bioinformatics or medical science. Gradient boosting decision tree (GBDT) is typically optimal for small datasets, while deep learning often performs better for large datasets. This work reports a boosting tree-assisted multitask deep learning (BTAMDL) architecture that integrates GBDT and multitask deep learning (MDL) to achieve near-optimal predictions for small datasets when there exists a large dataset that is well correlated to the small datasets. Two BTAMDL models are constructed, one utilizing purely MDL output as GBDT input while the other admitting additional features in GBDT input. The proposed BTAMDL models are validated on four categories of datasets, including toxicity, partition coefficient, solubility, and solvation. It is found that the proposed BTAMDL models outperform the current state-of-the-art methods in various applications involving small datasets.

Analysis of substance use and its outcomes by machine learning I. Childhood evaluation of liability to substance use disorder

BACKGROUND

Substance use disorder (SUD) exacts enormous societal costs in the United States, and it is important to detect high-risk youths for prevention. Machine learning (ML) is the method to find patterns and make prediction from data. We hypothesized that ML identifies the health, psychological, psychiatric, and contextual features to predict SUD, and the identified features predict high-risk individuals to develop SUD.

METHOD

Male (N = 494) and female (N = 206) participants and their informant parents were administered a battery of questionnaires across five waves of assessment conducted at 10-12, 12-14, 16, 19, and 22 years of age. Characteristics most strongly associated with SUD were identified using the random forest (RF)algorithm from approximately 1000 variables measured at each assessment. Next, the complement of features was validated, and the best models were selected for predicting SUD using seven ML algorithms. Lastly, area under the receiver operating characteristic curve (AUROC) evaluated accuracy of detecting individuals who develop SUD+/- up to thirty years of age.

RESULTS

Approximately thirty variables strongly predict SUD. The predictors shift from psychological dysregulation and poor health behavior in late childhood to non-normative socialization in mid to late adolescence. In 10-12-year-old youths, the features predict SUD+/- with 74% accuracy, increasing to 86% at 22 years of age. The RF algorithm optimally detects individuals between 10-22 years of age who develop SUD compared to other ML algorithms.

CONCLUSION

These findings inform the items required for inclusion in instruments to accurately identify high risk youths and young adults requiring SUD prevention.

General and efficient synthesis of 1,2-dihydropyrrolo[3,4-b]indol-3-ones via a formal [3 + 2] cycloaddition initiated by C–H activation

A [Cp*RhCl₂]₂-catalyzed formal [3 + 2] cycloaddition involving a sequential coupling reaction initiated by C–H activation and aza-Michael addition has been developed for the general and efficient synthesis of 1,2-dihydropyrrolo[3,4-b]indol-3-ones.

Rhodium(III)-Catalyzed Oxidative [3 + 2] Annulation of 2-Acetyl-1-arylhydrazines with Maleimides: Synthesis of Pyrrolo[3,4-b]indole-1,3-diones

A rhodium-catalyzed oxidative [3 + 2] annulation of 2-acetyl-1-phenylhydrazines with maleimides was accomplished using AgNTf₂ and Ag₂CO₃ as additive and oxidant, respectively. A variety of 2-acetyl-1-phenylhydrazines with maleimides were converted into pyrrolo[3,4-b]indole-1,3-diones in satisfactory to excellent yields. Synthetically useful functional groups, such as halogen atoms (F, Cl, Br, and I), ester, cyano, and nitro, remained intact during tandem C-H activation and annulation reactions.

Three-Dimensional Analysis of Binding Sites for Predicting Binding Affinities in Drug Design

Understanding the interaction between drug molecules and proteins is one of the main challenges in drug design. Several tools have been developed recently to decrease the complexity of the process. Artificial intelligence and machine learning methods offer promising results in predicting the binding affinities. It becomes possible to do accurate predictions by using the known protein-ligand interactions. In this study, the electrostatic potential values extracted from 3-dimensional grid cubes of the drug-protein binding sites are used for predicting binding affinities of related complexes. A new algorithm with a dynamic feature selection method was implemented, which is derived from Compressed Images For Affinity Prediction (CIFAP) study, to predict binding affinities of Checkpoint Kinase 1 and Caspase 3 inhibitors.

Concepts of Artificial Intelligence for Computer-Assisted Drug Discovery

Artificial intelligence (AI), and, in particular, deep learning as a subcategory of AI, provides opportunities for the discovery and development of innovative drugs. Various machine learning approaches have recently (re)emerged, some of which may be considered instances of domain-specific AI which have been successfully employed for drug discovery and design. This review provides a comprehensive portrayal of these machine learning techniques and of their applications in medicinal chemistry. After introducing the basic principles, alongside some application notes, of the various machine learning algorithms, the current state-of-the art of AI-assisted pharmaceutical discovery is discussed, including applications in structure- and ligand-based virtual screening, de novo drug design, physicochemical and pharmacokinetic property prediction, drug repurposing, and related aspects. Finally, several challenges and limitations of the current methods are summarized, with a view to potential future directions for AI-assisted drug discovery and design.

Prediction of Orthosteric and Allosteric Regulations on Cannabinoid Receptors Using Supervised Machine Learning Classifiers

Designing highly selective compounds to protein subtypes and developing allosteric modulators targeting them are critical considerations to both drug discovery and mechanism studies for cannabinoid receptors. It is challenging but in demand to have classifiers to identify active ligands from inactive or random compounds and distinguish allosteric modulators from orthosteric ligands. In this study, supervised machine learning classifiers were built for two subtypes of cannabinoid receptors, CB1 and CB2. Three types of features, including molecular descriptors, MACCS fingerprints, and ECFP6 fingerprints, were calculated to evaluate the compound sets from diverse aspects. Deep neural networks, as well as conventional machine learning algorithms including support vector machine, naïve Bayes, logistic regression, and ensemble learning, were applied. Their performances on the classification with different types of features were compared and discussed. According to the receiver operating characteristic curves and the calculated metrics, the advantages and drawbacks of each algorithm were investigated. The feature ranking was followed to help extract useful knowledge about critical molecular properties, substructural keys, and circular fingerprints. The extracted features will then facilitate the research on cannabinoid receptors by providing guidance on preferred properties for compound modification and novel scaffold design. Besides using conventional molecular docking studies for compound virtual screening, machine-learning-based decision-making models provide alternative options. This study can be of value to the application of machine learning in the area of drug discovery and compound development.