Machine Learning-Based Virtual Screening and Molecular Simulation Approaches Identified Novel Potential Inhibitors for Cancer Therapy

Combining machine learning, molecular dynamics, and free energy analysis for (5HT)-2A receptor modulator classification

The 5-Hydroxytryptamine (5HT)-2A receptor, a key target in psychoactive drug development, presents significant challenges in the design of selective compounds. Here, we describe the construction, evaluation and validation of two machine learning (ML) models for the classification of bioactivity mechanisms against the (5HT)-2A receptor. Employing neural networks and XGBoost models, we achieved an overall accuracy of around 87 %, which was further enhanced through molecular modelling (MM) (e.g. molecular dynamics simulations) and binding free energy analysis. This ML-MM integration provided insights into the mechanisms of direct modulators and prodrugs. A significant outcome of the current study is the development of a 'binding free energy fingerprint' specific to (5HT)-2A modulators, offering a novel metric for evaluating drug efficacy against this target. Our study demonstrates the prospective of employing a successful workflow combining AI with structural biology, offering a powerful tool for advancing psychoactive drug discovery.

Discovering potential asthma therapeutics targeting hematopoietic prostaglandin D2 synthase: An integrated computational approach

Allergic asthma, a chronic inflammatory illness that affects millions worldwide, has serious economic and health consequences. Despite advances in therapy, contemporary treatments have poor efficacy and negative effects. This study investigates hematopoietic prostaglandin D2 synthase (HPGDS) as a potential target for novel asthma therapies. Targeting HPGDS may provide innovative treatment methods. A library of phytochemicals was used to find putative HPGDS inhibitors by structure-based and ligand-based virtual screening. Among the 2295 compounds screened, four compounds (ZINC208828240, ZINC95627530, ZINC14727536, and ZINC14711790) demonstrated strong binding affinities of -10.4, -10.3, -9.2, -9.1 kcal/mol respectively with key residues, suggesting their potential as a highly effective HPGDS inhibitor. Molecular dynamics (MD) simulations and Molecular Mechanics Poisson-Boltzmann Surface Area (MMPBSA) computations were further performed to evaluate the stability and binding affinity of the complexes. MD simulations and MMPBSA confirmed that compound ZINC14711790 showed high stability and binding affinity (binding energy -31.52 kcal/mol) than other compounds, including HQL-79, suggesting that this compound might be used as promising inhibitors to treat asthma. RMSD and RMSF analysis also revealed that ZINC14711790 exhibited strong dynamic stability. The findings of this study show the efficacy of ZINC14711790 as HPGDS inhibitors with high binding affinity, dynamic stability, and appropriate ADMET profile.

Computational insights into allosteric inhibition of focal adhesion kinase: A combined pharmacophore modeling and molecular dynamics approach

Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase that modulates integrin and growth factor signaling pathways and is implicated in cancer cell migration, proliferation, and survival. Over the past decade various, FAK kinase, FERM, and FAT domain inhibitors have been reported and a few kinase domain inhibitors are under clinical consideration. However, few of them were identified as multikinase inhibitors. In kinase drug design selectivity is always a point of concern, to improve selectivity allosteric inhibitor development is the best choice. The current research utilized a pharmacophore modeling (PM) approach to identify novel allosteric inhibitors of FAK. The all-available allosteric inhibitor bound 3D structures with PDB ids 4EBV, 4EBW, and 4I4F were utilized for the pharmacophore modeling. The validated PM models were utilized to map a database of 770,550 compounds prepared from ZINC, EXIMED, SPECS, ASINEX, and InterBioScreen, aiming to identify potential allosteric inhibitors. The obtained compounds from screening step were forwarded to molecular docking (MD) for the prediction of binding orientation inside the allosteric site and the results were evaluated with the known FAK allosteric inhibitor (REF). Finally, 14 FAK-inhibitor complexes were selected from the docking study and were studied under molecular dynamics simulations (MDS) for 500 ns. The complexes were ranked according to binding free energy (BFE) and those demonstrated higher affinity for allosteric site of FAK than REF inhibitors were selected. The selected complexes were further analyzed for intermolecular interactions and finally, three potential allosteric inhibitor candidates for the inhibition of FAK protein were identified. We believe that identified scaffolds may help in drug development against FAK as an anticancer agent.

Leveraging shape screening and molecular dynamics simulations to optimize PARP1-Specific chemo/radio-potentiators for antitumor drug design

PARP1 plays a pivotal role in DNA repair within the base excision pathway, making it a promising therapeutic target for cancers involving BRCA mutations. Current study is focused on the discovery of PARP inhibitors with enhanced selectivity for PARP1. Concurrent inhibition of PARP1 with PARP2 and PARP3 affects cellular functions, potentially causing DNA damage accumulation and disrupting immune responses. In step 1, a virtual library of 593 million compounds has been screened using a shape-based screening approach to narrow down the promising scaffolds. In step 2, hierarchical docking approach embedded in Schrödinger suite was employed to select compounds with good dock score, drug-likeness and MMGBSA score. Analysis supplemented with decomposition energy, molecular dynamics (MD) simulations and hydrogen bond frequency analysis, pinpointed that active site residues; H862, G863, R878, M890, Y896 and F897 are crucial for specific binding of ZINC001258189808 and ZINC000092332196 with PARP1 as compared to PARP2 and PARP3. The binding of ZINC000656130962, ZINC000762230673, ZINC001332491123, and ZINC000579446675 also revealed interaction involving two additional active site residues of PARP1, namely N767 and E988. Weaker or no interaction was observed for these residues with PARP2 and PARP3. This approach advances our understanding of PARP-1 specific inhibitors and their mechanisms of action, facilitating the development of targeted therapeutics.

Triple Generative Self-Supervised Learning Method for Molecular Property Prediction

Molecular property prediction is an important task in drug discovery, and with help of self-supervised learning methods, the performance of molecular property prediction could be improved by utilizing large-scale unlabeled dataset. In this paper, we propose a triple generative self-supervised learning method for molecular property prediction, called TGSS. Three encoders including a bi-directional long short-term memory recurrent neural network (BiLSTM), a Transformer, and a graph attention network (GAT) are used in pre-training the model using molecular sequence and graph structure data to extract molecular features. The variational auto encoder (VAE) is used for reconstructing features from the three models. In the downstream task, in order to balance the information between different molecular features, a feature fusion module is added to assign different weights to each feature. In addition, to improve the interpretability of the model, atomic similarity heat maps were introduced to demonstrate the effectiveness and rationality of molecular feature extraction. We demonstrate the accuracy of the proposed method on chemical and biological benchmark datasets by comparative experiments.

Therapeutic Decision Making in Prevascular Mediastinal Tumors Using CT Radiomics and Clinical Features: Upfront Surgery or Pretreatment Needle Biopsy?

The study aimed to develop machine learning (ML) classification models for differentiating patients who needed direct surgery from patients who needed core needle biopsy among patients with prevascular mediastinal tumor (PMT). Patients with PMT who received a contrast-enhanced computed tomography (CECT) scan and initial management for PMT between January 2010 and December 2020 were included in this retrospective study. Fourteen ML algorithms were used to construct candidate classification models via the voting ensemble approach, based on preoperative clinical data and radiomic features extracted from the CECT. The classification accuracy of clinical diagnosis was 86.1%. The first ensemble learning model was built by randomly choosing seven ML models from a set of fourteen ML models and had a classification accuracy of 88.0% (95% CI = 85.8 to 90.3%). The second ensemble learning model was the combination of five ML models, including NeuralNetFastAI, NeuralNetTorch, RandomForest with Entropy, RandomForest with Gini, and XGBoost, and had a classification accuracy of 90.4% (95% CI = 87.9 to 93.0%), which significantly outperformed clinical diagnosis (p < 0.05). Due to the superior performance, the voting ensemble learning clinical-radiomic classification model may be used as a clinical decision support system to facilitate the selection of the initial management of PMT.

Application of Machine Learning Techniques to Assess Alpha-Fetoprotein at Diagnosis of Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the most common primary liver tumor and is associated with high mortality rates. Approximately 80% of cases occur in cirrhotic livers, posing a significant challenge for appropriate therapeutic management. Adequate screening programs in high-risk groups are essential for early-stage detection. The extent of extrahepatic tumor spread and hepatic functional reserve are recognized as two of the most influential prognostic factors. In this retrospective multicenter study, we utilized machine learning (ML) methods to analyze predictors of mortality at the time of diagnosis in a total of 208 patients. The eXtreme gradient boosting (XGB) method achieved the highest values in identifying key prognostic factors for HCC at diagnosis. The etiology of HCC was found to be the variable most strongly associated with a poorer prognosis. The widely used Barcelona Clinic Liver Cancer (BCLC) classification in our setting demonstrated superiority over the TNM classification. Although alpha-fetoprotein (AFP) remains the most commonly used biological marker, elevated levels did not correlate with reduced survival. Our findings suggest the need to explore new prognostic biomarkers for individualized management of these patients.

Extreme Gradient Boosting Combined with Conformal Predictors for Informative Solubility Estimation

We used the extreme gradient boosting (XGB) algorithm to predict the experimental solubility of chemical compounds in water and organic solvents and to select significant molecular descriptors. The accuracy of prediction of our forward stepwise top-importance XGB (FSTI-XGB) on curated solubility data sets in terms of RMSE was found to be 0.59-0.76 Log(S) for two water data sets, while for organic solvent data sets it was 0.69-0.79 Log(S) for the Methanol data set, 0.65-0.79 for the Ethanol data set, and 0.62-0.70 Log(S) for the Acetone data set. That was the first step. In the second step, we used uncurated and curated AquaSolDB data sets for applicability domain (AD) tests of Drugbank, PubChem, and COCONUT databases and determined that more than 95% of studied ca. 500,000 compounds were within the AD. In the third step, we applied conformal prediction to obtain narrow prediction intervals and we successfully validated them using test sets' true solubility values. With prediction intervals obtained in the last fourth step, we were able to estimate individual error margins and the accuracy class of the solubility prediction for molecules within the AD of three public databases. All that was possible without the knowledge of experimental database solubilities. We find these four steps novel because usually, solubility-related works only study the first step or the first two steps.

Prediction of Tribological Properties of UHMWPE/SiC Polymer Composites Using Machine Learning Techniques

Polymer composites are a class of material that are gaining a lot of attention in demanding tribological applications due to the ability of manipulating their performance by changing various factors, such as processing parameters, types of fillers, and operational parameters. Hence, a number of samples under different conditions need to be repeatedly produced and tested in order to satisfy the requirements of an application. However, with the advent of a new field of triboinformatics, which is a scientific discipline involving computer technology to collect, store, analyze, and evaluate tribological properties, we presently have access to a variety of high-end tools, such as various machine learning (ML) techniques, which can significantly aid in efficiently gauging the polymer's characteristics without the need to invest time and money in a physical experimentation. The development of an accurate model specifically for predicting the properties of the composite would not only cheapen the process of product testing, but also bolster the production rates of a very strong polymer combination. Hence, in the current study, the performance of five different machine learning (ML) techniques is evaluated for accurately predicting the tribological properties of ultrahigh molecular-weight polyethylene (UHMWPE) polymer composites reinforced with silicon carbide (SiC) nanoparticles. Three input parameters, namely, the applied pressure, holding time, and the concentration of SiCs, are considered with the specific wear rate (SWR) and coefficient of friction (COF) as the two output parameters. The five techniques used are support vector machines (SVMs), decision trees (DTs), random forests (RFs), k-nearest neighbors (KNNs), and artificial neural networks (ANNs). Three evaluation statistical metrics, namely, the coefficient of determination (R2-value), mean absolute error (MAE), and root mean square error (RMSE), are used to evaluate and compare the performances of the different ML techniques. Based upon the experimental dataset, the SVM technique was observed to yield the lowest error rates-with the RMSE being 2.09 × 10-4 and MAE being 2 × 10-4 for COF and for SWR, an RMSE of 2 × 10-4 and MAE of 1.6 × 10-4 were obtained-and highest R2-values of 0.9999 for COF and 0.9998 for SWR. The observed performance metrics shows the SVM as the most reliable technique in predicting the tribological properties-with an accuracy of 99.99% for COF and 99.98% for SWR-of the polymer composites.

A High-Generalizability Machine Learning Framework for Analyzing the Homogenized Properties of Short Fiber-Reinforced Polymer Composites

This study aims to develop a high-generalizability machine learning framework for predicting the homogenized mechanical properties of short fiber-reinforced polymer composites. The ensemble machine learning model (EML) employs a stacking algorithm using three base models of Extra Trees (ET), eXtreme Gradient Boosting machine (XGBoost), and Light Gradient Boosting machine (LGBM). A micromechanical model of a two-step homogenization algorithm is adopted and verified as an effective approach to composite modeling with randomly distributed fibers, which is integrated with finite element simulations for providing a high-quality ground-truth dataset. The model performance is thoroughly assessed for its accuracy, efficiency, interpretability, and generalizability. The results suggest that: (1) the EML model outperforms the base members on prediction accuracy, achieving R2 values of 0.988 and 0.952 on the train and test datasets, respectively; (2) the SHapley Additive exPlanations (SHAP) analysis identifies the Young's modulus of matrix, fiber, and fiber content as the top three factors influencing the homogenized properties, whereas the anisotropy is predominantly determined by the fiber orientations; (3) the EML model showcases good generalization capability on experimental data, and it has been shown to be more effective than high-fidelity computational models by significantly lowering computational costs while maintaining high accuracy.