SynerGNet: A Graph Neural Network Model to Predict Anticancer Drug Synergy.

Drug combination therapy shows promise in cancer treatment by addressing drug resistance, reducing toxicity, and enhancing therapeutic efficacy. However, the intricate and dynamic nature of biological systems makes identifying potential synergistic drugs a costly and time-consuming endeavor. To facilitate the development of combination therapy, techniques employing artificial intelligence have emerged as a transformative solution, providing a sophisticated avenue for advancing existing therapeutic approaches. In this study, we developed SynerGNet, a graph neural network model designed to accurately predict the synergistic effect of drug pairs against cancer cell lines. SynerGNet utilizes cancer-specific featured graphs created by integrating heterogeneous biological features into the human protein-protein interaction network, followed by a reduction process to enhance topological diversity. Leveraging synergy data provided by AZ-DREAM Challenges, the model yields a balanced accuracy of 0.68, significantly outperforming traditional machine learning. Encouragingly, augmenting the training data with carefully constructed synthetic instances improved the balanced accuracy of SynerGNet to 0.73. Finally, the results of an independent validation conducted against DrugCombDB demonstrated that it exhibits a strong performance when applied to unseen data. SynerGNet shows a great potential in detecting drug synergy, positioning itself as a valuable tool that could contribute to the advancement of combination therapy for cancer treatment.

Augmented drug combination dataset to improve the performance of machine learning models predicting synergistic anticancer effects.

Combination therapy has gained popularity in cancer treatment as it enhances the treatment efficacy and overcomes drug resistance. Although machine learning (ML) techniques have become an indispensable tool for discovering new drug combinations, the data on drug combination therapy currently available may be insufficient to build high-precision models. We developed a data augmentation protocol to unbiasedly scale up the existing anti-cancer drug synergy dataset. Using a new drug similarity metric, we augmented the synergy data by substituting a compound in a drug combination instance with another molecule that exhibits highly similar pharmacological effects. Using this protocol, we were able to upscale the AZ-DREAM Challenges dataset from 8798 to 6,016,697 drug combinations. Comprehensive performance evaluations show that ML models trained on the augmented data consistently achieve higher accuracy than those trained solely on the original dataset. Our data augmentation protocol provides a systematic and unbiased approach to generating more diverse and larger-scale drug combination datasets, enabling the development of more precise and effective ML models. The protocol presented in this study could serve as a foundation for future research aimed at discovering novel and effective drug combinations for cancer treatment.

Augmented drug combination dataset to improve the performance of machine learning models predicting synergistic anticancer effects.

Combination therapy has gained popularity in cancer treatment as it enhances the treatment efficacy and overcomes drug resistance. Although machine learning (ML) techniques have become an indispensable tool for discovering new drug combinations, the data on drug combination therapy currently available may be insufficient to build high-precision models. We developed a data augmentation protocol to unbiasedly scale up the existing anti-cancer drug synergy dataset. Using a new drug similarity metric, we augmented the synergy data by substituting a compound in a drug combination instance with another molecule that exhibits highly similar pharmacological effects. Using this protocol, we were able to upscale the AZ-DREAM Challenges dataset from 8,798 to 6,016,697 drug combinations. Comprehensive performance evaluations show that Random Forest and Gradient Boosting Trees models trained on the augmented data achieve higher accuracy than those trained solely on the original dataset. Our data augmentation protocol provides a systematic and unbiased approach to generating more diverse and larger-scale drug combination datasets, enabling the development of more precise and effective ML models. The protocol presented in this study could serve as a foundation for future research aimed at discovering novel and effective drug combinations for cancer treatment.

Unlocking the Potential of Kinase Targets in Cancer: Insights from CancerOmicsNet, an AI-Driven Approach to Drug Response Prediction in Cancer.

Deregulated protein kinases are crucial in promoting cancer cell proliferation and driving malignant cell signaling. Although these kinases are essential targets for cancer therapy due to their involvement in cell development and proliferation, only a small part of the human kinome has been targeted by drugs. A comprehensive scoring system is needed to evaluate and prioritize clinically relevant kinases. We recently developed CancerOmicsNet, an artificial intelligence model employing graph-based algorithms to predict the cancer cell response to treatment with kinase inhibitors. The performance of this approach has been evaluated in large-scale benchmarking calculations, followed by the experimental validation of selected predictions against several cancer types. To shed light on the decision-making process of CancerOmicsNet and to better understand the role of each kinase in the model, we employed a customized saliency map with adjustable channel weights. The saliency map, functioning as an explainable AI tool, allows for the analysis of input contributions to the output of a trained deep-learning model and facilitates the identification of essential kinases involved in tumor progression. The comprehensive survey of biomedical literature for essential kinases selected by CancerOmicsNet demonstrated that it could help pinpoint potential druggable targets for further investigation in diverse cancer types.

Identification of oral therapeutics using an AI platform against the virus responsible for COVID-19, SARS-CoV-2.

Purpose: This study introduces a sophisticated computational pipeline, eVir, designed for the discovery of antiviral drugs based on their interactions within the human protein network. There is a pressing need for cost-effective therapeutics for infectious diseases (e.g., COVID-19), particularly in resource-limited countries. Therefore, our team devised an Artificial Intelligence (AI) system to explore repurposing opportunities for currently used oral therapies. The eVir system operates by identifying pharmaceutical compounds that mirror the effects of antiviral peptides (AVPs)-fragments of human proteins known to interfere with fundamental phases of the viral life cycle: entry, fusion, and replication. eVir extrapolates the probable antiviral efficacy of a given compound by analyzing its established and predicted impacts on the human protein-protein interaction network. This innovative approach provides a promising platform for drug repurposing against SARS-CoV-2 or any virus for which peptide data is available. Methods: The eVir AI software pipeline processes drug-protein and protein-protein interaction networks generated from open-source datasets. eVir uses Node2Vec, a graph embedding technique, to understand the nuanced connections among drugs and proteins. The embeddings are input a Siamese Network (SNet) and MLPs, each tailored for the specific mechanisms of entry, fusion, and replication, to evaluate the similarity between drugs and AVPs. Scores generated from the SNet and MLPs undergo a Platt probability calibration and are combined into a unified score that gauges the potential antiviral efficacy of a drug. This integrated approach seeks to boost drug identification confidence, offering a potential solution for detecting therapeutic candidates with pronounced antiviral potency. Once identified a number of compounds were tested for efficacy and toxicity in lung carcinoma cells (Calu-3) infected with SARS-CoV-2. A lead compound was further identified to determine its efficacy and toxicity in K18-hACE2 mice infected with SARS-CoV-2. Computational Predictions: The SNet confidently differentiated between similar and dissimilar drug pairs with an accuracy of 97.28% and AUC of 99.47%. Key compounds identified through these networks included Zinc, Mebendazole, Levomenol, Gefitinib, Niclosamide, and Imatinib. Notably, Mebendazole and Zinc showcased the highest similarity scores, while Imatinib, Levemenol, and Gefitinib also ranked within the top 20, suggesting their significant pharmacological potentials. Further examination of protein binding analysis using explainable AI focused on reverse engineering the causality of the networks. Protein interaction scores for Mebendazole and Imatinib revealed their effects on notable proteins such as CDPK1, VEGF2, ABL1, and several tyrosine protein kinases. Laboratory Studies: This study determined that Mebendazole, Gefitinib, Topotecan and to some extent Carfilzomib showed conventional drug-response curves, with IC50 values near or below that of Remdesivir with excellent confidence all above R2>0.91, and no cytotoxicity at the IC50 concentration in Calu-3 cells. Cyclosporine A showed antiviral activity, but also unconventional drug-response curves and low R2 which are explained by the non-dose dependent toxicity of the compound. Additionally, Niclosamide demonstrated a conventional drug-response curve with high confidence; however, its inherent cytotoxicity may be a confounding element that misrepresents true antiviral efficacy, by reflecting cellular damage rather than a genuine antiviral action. Remdesivir was used as a control compound and was evaluated in parallel with the submitted test article and had conventional drug-response curves validating the overall results of the assay. Mebendazole was identified from the cell studies to have efficacy at non-toxic concentrations and were further evaluated in mice infected with SARS-CoV-2. Mebendazole administered to K18-hACE2 mice infected with SARS-CoV-2, resulted in a 44.2% reduction in lung viral load compared to non-treated placebo control respectively. There were no significant differences in body weight and all clinical chemistry determinations evaluated (i.e., kidney and liver enzymes) between the different treatment groups. Conclusion: This research underscores the potential of repurposing existing compounds for treating COVID-19. Our preliminary findings underscore the therapeutic promise of several compounds, notably Mebendazole, in both in vitro and in vivo settings against SARS-CoV-2. Several of the drugs explored, especially Mebendazole, are off-label medication; their cost-effectiveness position them as economical therapies against SARS-CoV-2.

Artificial intelligence to guide precision anticancer therapy with multitargeted kinase inhibitors.

BACKGROUND: Vast amounts of rapidly accumulating biological data related to cancer and a remarkable progress in the field of artificial intelligence (AI) have paved the way for precision oncology. Our recent contribution to this area of research is CancerOmicsNet, an AI-based system to predict the therapeutic effects of multitargeted kinase inhibitors across various cancers. This approach was previously demonstrated to outperform other deep learning methods, graph kernel models, molecular docking, and drug binding pocket matching. METHODS: CancerOmicsNet integrates multiple heterogeneous data by utilizing a deep graph learning model with sophisticated attention propagation mechanisms to extract highly predictive features from cancer-specific networks. The AI-based system was devised to provide more accurate and robust predictions than data-driven therapeutic discovery using gene signature reversion. RESULTS: Selected CancerOmicsNet predictions obtained for "unseen" data are positively validated against the biomedical literature and by live-cell time course inhibition assays performed against breast, pancreatic, and prostate cancer cell lines. Encouragingly, six molecules exhibited dose-dependent antiproliferative activities, with pan-CDK inhibitor JNJ-7706621 and Src inhibitor PP1 being the most potent against the pancreatic cancer cell line Panc 04.03. CONCLUSIONS: CancerOmicsNet is a promising AI-based platform to help guide the development of new approaches in precision oncology involving a variety of tumor types and therapeutics.

GraphSite: Ligand Binding Site Classification with Deep Graph Learning.

The binding of small organic molecules to protein targets is fundamental to a wide array of cellular functions. It is also routinely exploited to develop new therapeutic strategies against a variety of diseases. On that account, the ability to effectively detect and classify ligand binding sites in proteins is of paramount importance to modern structure-based drug discovery. These complex and non-trivial tasks require sophisticated algorithms from the field of artificial intelligence to achieve a high prediction accuracy. In this communication, we describe GraphSite, a deep learning-based method utilizing a graph representation of local protein structures and a state-of-the-art graph neural network to classify ligand binding sites. Using neural weighted message passing layers to effectively capture the structural, physicochemical, and evolutionary characteristics of binding pockets mitigates model overfitting and improves the classification accuracy. Indeed, comprehensive cross-validation benchmarks against a large dataset of binding pockets belonging to 14 diverse functional classes demonstrate that GraphSite yields the class-weighted F1-score of 81.7%, outperforming other approaches such as molecular docking and binding site matching. Further, it also generalizes well to unseen data with the F1-score of 70.7%, which is the expected performance in real-world applications. We also discuss new directions to improve and extend GraphSite in the future.

CancerOmicsNet: a multi-omics network-based approach to anti-cancer drug profiling.

Development of novel anti-cancer treatments requires not only a comprehensive knowledge of cancer processes and drug mechanisms of action, but also the ability to accurately predict the response of various cancer cell lines to therapeutics. Numerous computational methods have been developed to address this issue, including algorithms employing supervised machine learning. Nonetheless, high prediction accuracies reported for many of these techniques may result from a significant overlap among training, validation, and testing sets, making existing predictors inapplicable to new data. To address these issues, we developed CancerOmicsNet, a graph neural network with sophisticated attention propagation mechanisms to predict the therapeutic effects of kinase inhibitors across various tumors. Emphasizing on the system-level complexity of cancer, CancerOmicsNet integrates multiple heterogeneous data, such as biological networks, genomics, inhibitor profiling, and gene-disease associations, into a unified graph structure. The performance of CancerOmicsNet, properly cross-validated at the tissue level, is 0.83 in terms of the area under the receiver operating characteristics, which is notably higher than those measured for other approaches. CancerOmicsNet generalizes well to unseen data, i.e., it can predict therapeutic effects across a variety of cancer cell lines and inhibitors. CancerOmicsNet is freely available to the academic community at https://github.com/pulimeng/CancerOmicsNet.

An integrated network representation of multiple cancer-specific data for graph-based machine learning.

Genomic profiles of cancer cells provide valuable information on genetic alterations in cancer. Several recent studies employed these data to predict the response of cancer cell lines to drug treatment. Nonetheless, due to the multifactorial phenotypes and intricate mechanisms of cancer, the accurate prediction of the effect of pharmacotherapy on a specific cell line based on the genetic information alone is problematic. Emphasizing on the system-level complexity of cancer, we devised a procedure to integrate multiple heterogeneous data, including biological networks, genomics, inhibitor profiling, and gene-disease associations, into a unified graph structure. In order to construct compact, yet information-rich cancer-specific networks, we developed a novel graph reduction algorithm. Driven by not only the topological information, but also the biological knowledge, the graph reduction increases the feature-only entropy while preserving the valuable graph-feature information. Subsequent comparative benchmarking simulations employing a tissue level cross-validation protocol demonstrate that the accuracy of a graph-based predictor of the drug efficacy is 0.68, which is notably higher than those measured for more traditional, matrix-based techniques on the same data. Overall, the non-Euclidean representation of the cancer-specific data improves the performance of machine learning to predict the response of cancer to pharmacotherapy. The generated data are freely available to the academic community at https://osf.io/dzx7b/ .

Pocket2Drug: An Encoder-Decoder Deep Neural Network for the Target-Based Drug Design.

Computational modeling is an essential component of modern drug discovery. One of its most important applications is to select promising drug candidates for pharmacologically relevant target proteins. Because of continuing advances in structural biology, putative binding sites for small organic molecules are being discovered in numerous proteins linked to various diseases. These valuable data offer new opportunities to build efficient computational models predicting binding molecules for target sites through the application of data mining and machine learning. In particular, deep neural networks are powerful techniques capable of learning from complex data in order to make informed drug binding predictions. In this communication, we describe Pocket2Drug, a deep graph neural network model to predict binding molecules for a given a ligand binding site. This approach first learns the conditional probability distribution of small molecules from a large dataset of pocket structures with supervised training, followed by the sampling of drug candidates from the trained model. Comprehensive benchmarking simulations show that using Pocket2Drug significantly improves the chances of finding molecules binding to target pockets compared to traditional drug selection procedures. Specifically, known binders are generated for as many as 80.5% of targets present in the testing set consisting of dissimilar data from that used to train the deep graph neural network model. Overall, Pocket2Drug is a promising computational approach to inform the discovery of novel biopharmaceuticals.

Exploring the effects of genetic variation on gene regulation in cancer in the context of 3D genome structure.

BACKGROUND: Numerous genome-wide association studies (GWAS) conducted to date revealed genetic variants associated with various diseases, including breast and prostate cancers. Despite the availability of these large-scale data, relatively few variants have been functionally characterized, mainly because the majority of single-nucleotide polymorphisms (SNPs) map to the non-coding regions of the human genome. The functional characterization of these non-coding variants and the identification of their target genes remain challenging. RESULTS: In this communication, we explore the potential functional mechanisms of non-coding SNPs by integrating GWAS with the high-resolution chromosome conformation capture (Hi-C) data for breast and prostate cancers. We show that more genetic variants map to regulatory elements through the 3D genome structure than the 1D linear genome lacking physical chromatin interactions. Importantly, the association of enhancers, transcription factors, and their target genes with breast and prostate cancers tends to be higher when these regulatory elements are mapped to high-risk SNPs through spatial interactions compared to simply using a linear proximity. Finally, we demonstrate that topologically associating domains (TADs) carrying high-risk SNPs also contain gene regulatory elements whose association with cancer is generally higher than those belonging to control TADs containing no high-risk variants. CONCLUSIONS: Our results suggest that many SNPs may contribute to the cancer development by affecting the expression of certain tumor-related genes through long-range chromatin interactions with gene regulatory elements. Integrating large-scale genetic datasets with the 3D genome structure offers an attractive and unique approach to systematically investigate the functional mechanisms of genetic variants in disease risk and progression.

Artificial intelligence for the discovery of novel antimicrobial agents for emerging infectious diseases.

The search for effective drugs to treat new and existing diseases is a laborious one requiring a large investment of capital, resources, and time. The coronavirus 2019 (COVID-19) pandemic has been a painful reminder of the lack of development of new antimicrobial agents to treat emerging infectious diseases. Artificial intelligence (AI) and other in silico techniques can drive a more efficient, cost-friendly approach to drug discovery by helping move potential candidates with better clinical tolerance forward in the pipeline. Several research teams have developed successful AI platforms for hit identification, lead generation, and lead optimization. In this review, we investigate the technologies at the forefront of spearheading an AI revolution in drug discovery and pharmaceutical sciences.

GraphDTI: A robust deep learning predictor of drug-target interactions from multiple heterogeneous data.

Traditional techniqueset identification, we developed GraphDTI, a robust machine learning framework integrating the molecular-level information on drugs, proteins, and binding sites with the system-level information on gene expression and protein-protein interactions. In order to properly evaluate the performance of GraphDTI, we compiled a high-quality benchmarking dataset and devised a new cluster-based cross-validation p to identify macromolecular targets for drugs utilize solely the information on a query drug and a putative target. Nonetheless, the mechanisms of action of many drugs depend not only on their binding affinity toward a single protein, but also on the signal transduction through cascades of molecular interactions leading to certain phenotypes. Although using protein-protein interaction networks and drug-perturbed gene expression profiles can facilitate system-level investigations of drug-target interactions, utilizing such large and heterogeneous data poses notable challenges. To improve the state-of-the-art in drug targrotocol. Encouragingly, GraphDTI not only yields an AUC of 0.996 against the validation dataset, but it also generalizes well to unseen data with an AUC of 0.939, significantly outperforming other predictors. Finally, selected examples of identified drug-target interactions are validated against the biomedical literature. Numerous applications of GraphDTI include the investigation of drug polypharmacological effects, side effects through off-target binding, and repositioning opportunities.

Bionoi: A Voronoi Diagram-Based Representation of Ligand-Binding Sites in Proteins for Machine Learning Applications.

Bionoi is a new software to generate Voronoi representations of ligand-binding sites in proteins for machine learning applications. Unlike many other deep learning models in biomedicine, Bionoi utilizes off-the-shelf convolutional neural network architectures, reducing the development work without sacrificing the performance. When initially generating images of binding sites, users have the option to color the Voronoi cells based on either one of six structural, physicochemical, and evolutionary properties, or a blend of all six individual properties. Encouragingly, after inputting images generated by Bionoi into the convolutional autoencoder, the network was able to effectively learn the most salient features of binding pockets. The accuracy of the generated model is evaluated both visually and numerically through the reconstruction of binding site images from the latent feature space. The generated feature vectors capture well various properties of binding sites and thus can be applied in a multitude of machine learning projects. As a demonstration, we trained the ResNet-18 architecture from Microsoft on Bionoi images to show that it is capable to effectively classify nucleotide- and heme-binding pockets against a large dataset of control pockets binding a variety of small molecules. Bionoi is freely available to the research community at https://github.com/CSBG-LSU/BionoiNet.

BionoiNet: ligand-binding site classification with off-the-shelf deep neural network.

MOTIVATION: Fast and accurate classification of ligand-binding sites in proteins with respect to the class of binding molecules is invaluable not only to the automatic functional annotation of large datasets of protein structures but also to projects in protein evolution, protein engineering and drug development. Deep learning techniques, which have already been successfully applied to address challenging problems across various fields, are inherently suitable to classify ligand-binding pockets. Our goal is to demonstrate that off-the-shelf deep learning models can be employed with minimum development effort to recognize nucleotide- and heme-binding sites with a comparable accuracy to highly specialized, voxel-based methods. RESULTS: We developed BionoiNet, a new deep learning-based framework implementing a popular ResNet model for image classification. BionoiNet first transforms the molecular structures of ligand-binding sites to 2D Voronoi diagrams, which are then used as the input to a pretrained convolutional neural network classifier. The ResNet model generalizes well to unseen data achieving the accuracy of 85.6% for nucleotide- and 91.3% for heme-binding pockets. BionoiNet also computes significance scores of pocket atoms, called BionoiScores, to provide meaningful insights into their interactions with ligand molecules. BionoiNet is a lightweight alternative to computationally expensive 3D architectures. AVAILABILITY AND IMPLEMENTATION: BionoiNet is implemented in Python with the source code freely available at: https://github.com/CSBG-LSU/BionoiNet. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Hinge Region in DNA Packaging Terminase pUL15 of Herpes Simplex Virus: A Potential Allosteric Target for Antiviral Drugs.

Approximately 80% of adults are infected with a member of the herpesviridae family. Herpesviruses establish life-long latent infections within neurons, which may reactivate into lytic infections due to stress or immune suppression. There are nine human herpesviruses (HHV) posing health concerns from benign conditions to life threatening encephalitis, including cancers associated with viral infections. The current treatment options for most HHV conditions mainly include several nucleoside and nucleotide analogs targeting viral DNA polymerase. Although these drugs help manage infections, their common mechanism of action may lead to the development of drug resistance, which is particularly devastating in immunocompromised patients. Therefore, new classes of drugs directed against novel targets in HHVs are necessary to alleviate this issue. We analyzed the conservation rates of all proteins in herpes simplex virus 1 (HHV-1), a representative of the HHV family and one of the most common viruses infecting the human population. Furthermore, we generated a full-length structure model of the most conserved HHV-1 protein, the DNA packaging terminase pUL15. A series of computational analyses were performed on the model to identify ATP and DNA binding sites and characterize the dynamics of the protein. Our study indicates that proteins involved in HHV-1 DNA packaging and cleavage are amongst the most conserved gene products of HHVs. Since the packaging protein pUL15 is the most conserved among all HHV-1 gene products, the virus will have a lower chance of developing resistance to small molecules targeting pUL15. A subsequent analysis of the structure of pUL15 revealed distinct ATP and DNA binding domains and the elastic network model identifies a functionally important hinge region between the two domains of pUL15. The atomic information on the active and allosteric sites in the ATP- and DNA-bound model of pUL15 presented in this study can inform the structure-based drug discovery of a new class of drugs to treat a wide range of HHVs.

Identification and Visualization of Functionally Important Domains and Residues in Herpes Simplex Virus Glycoprotein K(gK) Using a Combination of Phylogenetics and Protein Modeling.

Alphaherpesviruses are a subfamily of herpesviruses that include the significant human pathogens herpes simplex viruses (HSV) and varicella zoster virus (VZV). Glycoprotein K (gK), conserved in all alphaherpesviruses, is a multi-membrane spanning virion glycoprotein essential for virus entry into neuronal axons, virion assembly, and pathogenesis. Despite these critical functions, little is known about which gK domains and residues are most important for maintaining these functions across all alphaherpesviruses. Herein, we employed phylogenetic and structural analyses including the use of a novel model for evolutionary rate variation across residues to predict conserved gK functional domains. We found marked heterogeneity in the evolutionary rate at the level of both individual residues and domains, presumably as a result of varying selective constraints. To clarify the potential role of conserved sequence features, we predicted the structures of several gK orthologs. Congruent with our phylogenetic analysis, slowly evolving residues were identified at potentially structurally significant positions across domains. We found that using a quantitative measure of amino acid rate variation combined with molecular modeling we were able to identify amino acids predicted to be critical for gK protein structure/function. This analysis yields targets for the design of anti-herpesvirus therapeutic strategies across all alphaherpesvirus species that would be absent from more traditional analyses of conservation.

Elucidating the druggability of the human proteome with eFindSite.

Identifying the viability of protein targets is one of the preliminary steps of drug discovery. Determining the ability of a protein to bind drugs in order to modulate its function, termed the druggability, requires a non-trivial amount of time and resources. Inability to properly measure druggability has accounted for a significant portion of failures in drug discovery. This problem is only further exacerbated by the large sample space of proteins involved in human diseases. With these barriers, the druggability space within the human proteome remains unexplored and has made it difficult to develop drugs for numerous diseases. Hence, we present a new feature developed in eFindSite that employs supervised machine learning to predict the druggability of a given protein. Benchmarking calculations against the Non-Redundant data set of Druggable and Less Druggable binding sites demonstrate that an AUC for druggability prediction with eFindSite is as high as 0.88. With eFindSite, we elucidated the human druggability space to be 10,191 proteins. Considering the disease space from the Open Targets Platform and excluding already known targets from the predicted data set reveal 2731 potentially novel therapeutic targets. eFindSite is freely available as a stand-alone software at https://github.com/michal-brylinski/efindsite .

DeepDrug3D: Classification of ligand-binding pockets in proteins with a convolutional neural network.

Comprehensive characterization of ligand-binding sites is invaluable to infer molecular functions of hypothetical proteins, trace evolutionary relationships between proteins, engineer enzymes to achieve a desired substrate specificity, and develop drugs with improved selectivity profiles. These research efforts pose significant challenges owing to the fact that similar pockets are commonly observed across different folds, leading to the high degree of promiscuity of ligand-protein interactions at the system-level. On that account, novel algorithms to accurately classify binding sites are needed. Deep learning is attracting a significant attention due to its successful applications in a wide range of disciplines. In this communication, we present DeepDrug3D, a new approach to characterize and classify binding pockets in proteins with deep learning. It employs a state-of-the-art convolutional neural network in which biomolecular structures are represented as voxels assigned interaction energy-based attributes. The current implementation of DeepDrug3D, trained to detect and classify nucleotide- and heme-binding sites, not only achieves a high accuracy of 95%, but also has the ability to generalize to unseen data as demonstrated for steroid-binding proteins and peptidase enzymes. Interestingly, the analysis of strongly discriminative regions of binding pockets reveals that this high classification accuracy arises from learning the patterns of specific molecular interactions, such as hydrogen bonds, aromatic and hydrophobic contacts. DeepDrug3D is available as an open-source program at https://github.com/pulimeng/DeepDrug3D with the accompanying TOUGH-C1 benchmarking dataset accessible from https://osf.io/enz69/.