Network-Based Methods for Prediction of Drug-Target Interactions

Drug-target interaction (DTI) is the basis of drug discovery. However, it is time-consuming and costly to determine DTIs experimentally. Over the past decade, various computational methods were proposed to predict potential DTIs with high efficiency and low costs. These methods can be roughly divided into several categories, such as molecular docking-based, pharmacophore-based, similarity-based, machine learning-based, and network-based methods. Among them, network-based methods, which do not rely on three-dimensional structures of targets and negative samples, have shown great advantages over the others. In this article, we focused on network-based methods for DTI prediction, in particular our network-based inference (NBI) methods that were derived from recommendation algorithms. We first introduced the methodologies and evaluation of network-based methods, and then the emphasis was put on their applications in a wide range of fields, including target prediction and elucidation of molecular mechanisms of therapeutic effects or safety problems. Finally, limitations and perspectives of network-based methods were discussed. In a word, network-based methods provide alternative tools for studies in drug repurposing, new drug discovery, systems pharmacology and systems toxicology.

DTI-CDF: a cascade deep forest model towards the prediction of drug-target interactions based on hybrid features

Drug-target interactions (DTIs) play a crucial role in target-based drug discovery and development. Computational prediction of DTIs can effectively complement experimental wet-lab techniques for the identification of DTIs, which are typically time- and resource-consuming. However, the performances of the current DTI prediction approaches suffer from a problem of low precision and high false-positive rate. In this study, we aim to develop a novel DTI prediction method for improving the prediction performance based on a cascade deep forest (CDF) model, named DTI-CDF, with multiple similarity-based features between drugs and the similarity-based features between target proteins extracted from the heterogeneous graph, which contains known DTIs. In the experiments, we built five replicates of 10-fold cross-validation under three different experimental settings of data sets, namely, corresponding DTI values of certain drugs (SD), targets (ST), or drug-target pairs (SP) in the training sets are missed but existed in the test sets. The experimental results demonstrate that our proposed approach DTI-CDF achieves a significantly higher performance than that of the traditional ensemble learning-based methods such as random forest and XGBoost, deep neural network, and the state-of-the-art methods such as DDR. Furthermore, there are 1352 newly predicted DTIs which are proved to be correct by KEGG and DrugBank databases. The data sets and source code are freely available at https://github.com//a96123155/DTI-CDF.

Deep Learning in Virtual Screening: Recent Applications and Developments

Drug discovery is a cost and time-intensive process that is often assisted by computational methods, such as virtual screening, to speed up and guide the design of new compounds. For many years, machine learning methods have been successfully applied in the context of computer-aided drug discovery. Recently, thanks to the rise of novel technologies as well as the increasing amount of available chemical and bioactivity data, deep learning has gained a tremendous impact in rational active compound discovery. Herein, recent applications and developments of machine learning, with a focus on deep learning, in virtual screening for active compound design are reviewed. This includes introducing different compound and protein encodings, deep learning techniques as well as frequently used bioactivity and benchmark data sets for model training and testing. Finally, the present state-of-the-art, including the current challenges and emerging problems, are examined and discussed.

Comparison Study of Computational Prediction Tools for Drug-Target Binding Affinities

The drug development is generally arduous, costly, and success rates are low. Thus, the identification of drug-target interactions (DTIs) has become a crucial step in early stages of drug discovery. Consequently, developing computational approaches capable of identifying potential DTIs with minimum error rate are increasingly being pursued. These computational approaches aim to narrow down the search space for novel DTIs and shed light on drug functioning context. Most methods developed to date use binary classification to predict if the interaction between a drug and its target exists or not. However, it is more informative but also more challenging to predict the strength of the binding between a drug and its target. If that strength is not sufficiently strong, such DTI may not be useful. Therefore, the methods developed to predict drug-target binding affinities (DTBA) are of great value. In this study, we provide a comprehensive overview of the existing methods that predict DTBA. We focus on the methods developed using artificial intelligence (AI), machine learning (ML), and deep learning (DL) approaches, as well as related benchmark datasets and databases. Furthermore, guidance and recommendations are provided that cover the gaps and directions of the upcoming work in this research area. To the best of our knowledge, this is the first comprehensive comparison analysis of tools focused on DTBA with reference to AI/ML/DL.

Drug-target interaction prediction via chemogenomic space: learning-based methods

INTRODUCTION

Identification of the interaction between drugs and target proteins is a crucial task in genomic drug discovery. The in silico prediction is an appropriate alternative for the laborious and costly experimental process of drug-target interaction prediction. Developing a variety of computational methods opens a new direction in analyzing and detecting new drug-target pairs.

AREAS COVERED

In this review, we will focus on chemogenomic methods which have established a learning framework for predicting drug-target interactions. Learning-based methods are classified into supervised and semi-supervised, and the supervised learning methods are studied as two separate parts including similarity-based methods and feature-based methods.

EXPERT OPINION

In spite of many improvements for pharmacology applications by learning-based methods, there are many over simplification settings in construction of predictive models that may lead to over-optimistic results on drug-target interaction prediction.

Deep drug-target binding affinity prediction with multiple attention blocks

Drug-target interaction (DTI) prediction has drawn increasing interest due to its substantial position in the drug discovery process. Many studies have introduced computational models to treat DTI prediction as a regression task, which directly predict the binding affinity of drug-target pairs. However, existing studies (i) ignore the essential correlations between atoms when encoding drug compounds and (ii) model the interaction of drug-target pairs simply by concatenation. Based on those observations, in this study, we propose an end-to-end model with multiple attention blocks to predict the binding affinity scores of drug-target pairs. Our proposed model offers the abilities to (i) encode the correlations between atoms by a relation-aware self-attention block and (ii) model the interaction of drug representations and target representations by the multi-head attention block. Experimental results of DTI prediction on two benchmark datasets show our approach outperforms existing methods, which are benefit from the correlation information encoded by the relation-aware self-attention block and the interaction information extracted by the multi-head attention block. Moreover, we conduct the experiments on the effects of max relative position length and find out the best max relative position length value \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$k \in \{3, 5\}$\end{document}. Furthermore, we apply our model to predict the binding affinity of Corona Virus Disease 2019 (COVID-19)-related genome sequences and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$3137$\end{document} FDA-approved drugs.

SPVec: A Word2vec-Inspired Feature Representation Method for Drug-Target Interaction Prediction

Drug discovery is an academical and commercial process of global importance. Accurate identification of drug-target interactions (DTIs) can significantly facilitate the drug discovery process. Compared to the costly, labor-intensive and time-consuming experimental methods, machine learning (ML) plays an ever-increasingly important role in effective, efficient and high-throughput identification of DTIs. However, upstream feature extraction methods require tremendous human resources and expert insights, which limits the application of ML approaches. Inspired by the unsupervised representation learning methods like Word2vec, we here proposed SPVec, a novel way to automatically represent raw data such as SMILES strings and protein sequences into continuous, information-rich and lower-dimensional vectors, so as to avoid the sparseness and bit collisions from the cumbersomely manually extracted features. Visualization of SPVec nicely illustrated that the similar compounds or proteins occupy similar vector space, which indicated that SPVec not only encodes compound substructures or protein sequences efficiently, but also implicitly reveals some important biophysical and biochemical patterns. Compared with manually-designed features like MACCS fingerprints and amino acid composition (AAC), SPVec showed better performance with several state-of-art machine learning classifiers such as Gradient Boosting Decision Tree, Random Forest and Deep Neural Network on BindingDB. The performance and robustness of SPVec were also confirmed on independent test sets obtained from DrugBank database. Also, based on the whole DrugBank dataset, we predicted the possibilities of all unlabeled DTIs, where two of the top five predicted novel DTIs were supported by external evidences. These results indicated that SPVec can provide an effective and efficient way to discover reliable DTIs, which would be beneficial for drug reprofiling.

A heterogeneous network embedding framework for predicting similarity-based drug-target interactions

Accurate prediction of drug-target interactions (DTIs) through biological data can reduce the time and economic cost of drug development. The prediction method of DTIs based on a similarity network is attracting increasing attention. Currently, many studies have focused on predicting DTIs. However, such approaches do not consider the features of drugs and targets in multiple networks or how to extract and merge them. In this study, we proposed a Network EmbeDding framework in mulTiPlex networks (NEDTP) to predict DTIs. NEDTP builds a similarity network of nodes based on 15 heterogeneous information networks. Next, we applied a random walk to extract the topology information of each node in the network and learn it as a low-dimensional vector. Finally, the Gradient Boosting Decision Tree model was constructed to complete the classification task. NEDTP achieved accurate results in DTI prediction, showing clear advantages over several state-of-the-art algorithms. The prediction of new DTIs was also verified from multiple perspectives. In addition, this study also proposes a reasonable model for the widespread negative sampling problem of DTI prediction, contributing new ideas to future research. Code and data are available at https://github.com/LiangYu-Xidian/NEDTP.

Recent Advances in the Machine Learning-Based Drug-Target Interaction Prediction

BACKGROUND

The identification of drug-target interactions is a crucial issue in drug discovery. In recent years, researchers have made great efforts on the drug-target interaction predictions, and developed databases, software and computational methods.

RESULTS

In the paper, we review the recent advances in machine learning-based drug-target interaction prediction. First, we briefly introduce the datasets and data, and summarize features for drugs and targets which can be extracted from different data. Since drug-drug similarity and target-target similarity are important for many machine learning prediction models, we introduce how to calculate similarities based on data or features. Different machine learningbased drug-target interaction prediction methods can be proposed by using different features or information. Thus, we summarize, analyze and compare different machine learning-based prediction methods.

CONCLUSION

This study provides the guide to the development of computational methods for the drug-target interaction prediction.

Identification of key genes, regulatory factors, and drug target genes of recurrent implantation failure (RIF)

Objective: Recurrent implantation failure (RIF) exacerbates the physical trauma of infertile women that undergone in vitro fertilization-embryo transfer (IVF-ET). We aimed to identify the key genes, regulatory factors, and drug target genes involved in the RIF.Methods: The dataset GSE58144 that obtained from the Gene Expression Omnibus mainly contained 43 RIF and 72 control endometrial samples. Differently expressed genes (DEGs) between RIF and control groups were firstly analyzed, followed by the pathway and Gene Ontology (GO) enrichment analysis. Then, protein-protein interaction (PPI) network and miRNA-transcript factor (TF)-DEGs network were established. Finally, a drug-target interaction network was constructed.Results: A total of 399 DEGs were identified between the RIF and controls. In the PPI and key module network, UBE2I, PLK4, XPO1, AURKB, and NUP107 were identified as the hub genes, which mainly enriched in RNA transport and cell division cycle-related pathways and GO items. In the miRNA-TF-DEGs network, E2F4, SIN3A, miRNA489, miRNA199A, miRNA369-3P, miRNA422, and miRNA522 were considered as the key regulatory factors during RIF. In addition, HTR1A, NR3C1, and GABRA3 were the main targets of the drugs annotated in DrugBank.Conclusion: The effects of PLK4, XPO1, AURKB, and NUP107 on the RIF may be via affecting the proliferation and differentiation of endometrial stromal cells. Besides, SIN3A and miRNA199A may be crucial for embryo implantation. In addition, NR3C1 may be used as a possible target for the clinical therapy of RIF.

Pharmacokinetic Steady-States Highlight Interesting Target-Mediated Disposition Properties

In this paper, we derive explicit expressions for the concentrations of ligand L, target R and ligand-target complex RL at steady state for the classical model describing target-mediated drug disposition, in the presence of a constant-rate infusion of ligand. We demonstrate that graphing the steady-state values of ligand, target and ligand-target complex, we obtain striking and often singular patterns, which yield a great deal of insight and understanding about the underlying processes. Deriving explicit expressions for the dependence of L, R and RL on the infusion rate, and displaying graphs of the relations between L, R and RL, we give qualitative and quantitive information for the experimentalist about the processes involved. Understanding target turnover is pivotal for optimising these processes when target-mediated drug disposition (TMDD) prevails. By a combination of mathematical analysis and simulations, we also show that the evolution of the three concentration profiles towards their respective steady-states can be quite complex, especially for lower infusion rates. We also show how parameter estimates obtained from iv bolus studies can be used to derive steady-state concentrations of ligand, target and complex. The latter may serve as a template for future experimental designs.

DeepH-DTA: Deep Learning for Predicting Drug-Target Interactions: A Case Study of COVID-19 Drug Repurposing

The rapid spread of novel coronavirus pneumonia (COVID-19) has led to a dramatically increased mortality rate worldwide. Despite many efforts, the rapid development of an effective vaccine for this novel virus will take considerable time and relies on the identification of drug-target (DT) interactions utilizing commercially available medication to identify potential inhibitors. Motivated by this, we propose a new framework, called DeepH-DTA, for predicting DT binding affinities for heterogeneous drugs. We propose a heterogeneous graph attention (HGAT) model to learn topological information of compound molecules and bidirectional ConvLSTM layers for modeling spatio-sequential information in simplified molecular-input line-entry system (SMILES) sequences of drug data. For protein sequences, we propose a squeezed-excited dense convolutional network for learning hidden representations within amino acid sequences; while utilizing advanced embedding techniques for encoding both kinds of input sequences. The performance of DeepH-DTA is evaluated through extensive experiments against cutting-edge approaches utilising two public datasets (Davis, and KIBA) which comprise eclectic samples of the kinase protein family and the pertinent inhibitors. DeepH-DTA attains the highest Concordance Index (CI) of 0.924 and 0.927 and also achieved a mean square error (MSE) of 0.195 and 0.111 on the Davis and KIBA datasets respectively. Moreover, a study using FDA-approved drugs from the Drug Bank database is performed using DeepH-DTA to predict the affinity scores of drugs against SARS-CoV-2 amino acid sequences, and the results show that that the model can predict some of the SARS-Cov-2 inhibitors that have been recently approved in many clinical studies.

EmbedDTI: Enhancing the Molecular Representations via Sequence Embedding and Graph Convolutional Network for the Prediction of Drug-Target Interaction

The identification of drug-target interaction (DTI) plays a key role in drug discovery and development. Benefitting from large-scale drug databases and verified DTI relationships, a lot of machine-learning methods have been developed to predict DTIs. However, due to the difficulty in extracting useful information from molecules, the performance of these methods is limited by the representation of drugs and target proteins. This study proposes a new model called EmbedDTI to enhance the representation of both drugs and target proteins, and improve the performance of DTI prediction. For protein sequences, we leverage language modeling for pretraining the feature embeddings of amino acids and feed them to a convolutional neural network model for further representation learning. For drugs, we build two levels of graphs to represent compound structural information, namely the atom graph and substructure graph, and adopt graph convolutional network with an attention module to learn the embedding vectors for the graphs. We compare EmbedDTI with the existing DTI predictors on two benchmark datasets. The experimental results show that EmbedDTI outperforms the state-of-the-art models, and the attention module can identify the components crucial for DTIs in compounds.

A Multi-Label Learning Framework for Drug Repurposing

Drug repurposing plays an important role in screening old drugs for new therapeutic efficacy. The existing methods commonly treat prediction of drug-target interaction as a problem of binary classification, in which a large number of randomly sampled drug-target pairs accounting for over 50% of the entire training dataset are necessarily required. Such a large number of negative examples that do not come from experimental observations inevitably decrease the credibility of predictions. In this study, we propose a multi-label learning framework to find new uses for old drugs and discover new drugs for known target genes. In the framework, each drug is treated as a class label and its target genes are treated as the class-specific training data to train a supervised learning model of l₂-regularized logistic regression. As such, the inter-drug associations are explicitly modelled into the framework and all the class-specific training data come from experimental observations. In addition, the data constraint is less demanding, for instance, the chemical substructures of a drug are no longer needed and the novel target genes are inferred only from the underlying patterns of the known genes targeted by the drug. Stratified multi-label cross-validation shows that 84.9% of known target genes have at least one drug correctly recognized, and the proposed framework correctly recognizes 86.73% of the independent test drug-target interactions (DTIs) from DrugBank. These results show that the proposed framework could generalize well in the large drug/class space without the information of drug chemical structures and target protein structures. Furthermore, we use the trained model to predict new drugs for the known target genes, identify new genes for the old drugs, and infer new associations between old drugs and new disease phenotypes via the OMIM database. Gene ontology (GO) enrichment analyses and the disease associations reported in recent literature provide supporting evidences to the computational results, which potentially shed light on new clinical therapies for new and/or old disease phenotypes.

The Computational Models of Drug-target Interaction Prediction

The identification of Drug-Target Interactions (DTIs) is an important process in drug discovery and medical research. However, the tradition experimental methods for DTIs identification are still time consuming, extremely expensive and challenging. In the past ten years, various computational methods have been developed to identify potential DTIs. In this paper, the identification methods of DTIs are summarized. What's more, several state-of-the-art computational methods are mainly introduced, containing network-based method and machine learning-based method. In particular, for machine learning-based methods, including the supervised and semisupervised models, have essential differences in the approach of negative samples. Although these effective computational models in identification of DTIs have achieved significant improvements, network-based and machine learning-based methods have their disadvantages, respectively. These computational methods are evaluated on four benchmark data sets via values of Area Under the Precision Recall curve (AUPR).

Integrating Phenotypic Search and Phosphoproteomic Profiling of Active Kinases for Optimization of Drug Mixtures for RCC Treatment

Combined application of multiple therapeutic agents presents the possibility of enhanced efficacy and reduced development of resistance. Definition of the most appropriate combination for any given disease phenotype is challenged by the vast number of theoretically possible combinations of drugs and doses, making extensive empirical testing a virtually impossible task. We have used the streamlined-feedback system control (s-FSC) technique, a phenotypic approach, which converges to optimized drug combinations (ODC) within a few experimental steps. Phosphoproteomics analysis coupled to kinase activity analysis using the novel INKA (integrative inferred kinase activity) pipeline was performed to evaluate ODC mechanisms in a panel of renal cell carcinoma (RCC) cell lines. We identified different ODC with up to 95% effectivity for each RCC cell line, with low doses (ED_5-25) of individual drugs. Global phosphoproteomics analysis demonstrated inhibition of relevant kinases, and targeting remaining active kinases with additional compounds improved efficacy. In addition, we identified a common RCC ODC, based on kinase activity data, to be effective in all RCC cell lines under study. Combining s-FSC with a phosphoproteomic profiling approach provides valuable insight in targetable kinase activity and allows for the identification of superior drug combinations for the treatment of RCC.

Proteomic Profiling of Intra-Islet Features Reveals Substructure-Specific Protein Signatures

Despite their diminutive size, islets of Langerhans play a large role in maintaining systemic energy balance in the body. New technologies have enabled us to go from studying the whole pancreas to isolated whole islets, to partial islet sections, and now to islet substructures isolated from within the islet. Using a microfluidic nanodroplet-based proteomics platform coupled with laser capture microdissection and field asymmetric waveform ion mobility spectrometry, we present an in-depth investigation of protein profiles specific to features within the islet. These features include the islet-acinar interface vascular tissue, inner islet vasculature, isolated endocrine cells, whole islet with vasculature, and acinar tissue from around the islet. Compared to interface vasculature, unique protein signatures observed in the inner vasculature indicate increased innervation and intra-islet neuron-like crosstalk. We also demonstrate the utility of these data for identifying localized structure-specific drug-target interactions using existing protein/drug binding databases.

Identification of drug compounds for keloids and hypertrophic scars: drug discovery based on text mining and DeepPurpose

Background

Keloids (KL) and hypertrophic scars (HS) are forms of abnormal cutaneous scarring characterized by excessive deposition of extracellular matrix and fibroblast proliferation. Currently, the efficacy of drug therapies for KL and HS is limited. The present study aimed to investigate new drug therapies for KL and HS by using computational methods.

Methods

Text mining and GeneCodis were used to mine genes closely related to KL and HS. Protein-protein interaction analysis was performed using Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) and Cytoscape. The selection of drugs targeting the genes closely related to KL and HS was carried out using Pharmaprojects. Drug-target interaction prediction was performed using DeepPurpose, through which candidate drugs with the highest predicted binding affinity were finally obtained.

Results

Our analysis using text mining identified 69 KL- and HS-related genes. Gene enrichment analysis generated 25 genes, representing 7 pathways and 130 targeting drugs. DeepPurpose recommended 14 drugs as the final drug list, including 2 phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K) inhibitors, 10 prostaglandin-endoperoxide synthase 2 (PTGS2) inhibitors and 2 vascular endothelial growth factor A (VEGFA) antagonists.

Conclusions

Drug discovery using in silico text mining and DeepPurpose may be a powerful and effective way to identify drugs targeting the genes related to KL and HS.

Identification of potential drug-targets by combining evolutionary information extracted from frequency profiles and molecular topological structures

Identifying interactions among drug compounds and target proteins is the basis of drug research and plays a crucial role in drug discovery. However, determining drug-target interactions (DTIs) and potential protein-compound interactions by biological experiment-based method alone is a very complicated, expensive, and time-consuming process. Hence, there is an intense motivation to design in silico prediction methods to overcome these obstacles. In this work, we designed a novel in silico strategy to predict proteome-scale DTIs based on the assumption that DTI pairs can be expressed through the evolutionary information derived from frequency profiles and drugs' structural properties. To achieve this, drug molecules are encoded into the substructure fingerprints to represent certain fragments; target proteins are first converted into position-specific scoring matrix (PSSM) and then encoded as 2-dimensional principal component analysis (2DPCA) descriptors. In the prediction phase, the feature weighted rotation forest (RF) classifier is used to estimate whether drug and target interact with each other on four benchmark datasets, including Enzymes, Ion Channels, GPCRs, and Nuclear Receptors. The prediction accuracy of cross-validation on the four datasets is 95.40%, 88.82%, 85.67%, and 82.22%, respectively. In order to have a clearer assessment of the proposed approach, we compared it with the discrete cosine transform (DCT) descriptor model, support vector machine (SVM) classifier model, and existing excellent approaches, including DBSI, NetCBP, KBMF2K, SIMCOMP, and RFDT. The excellent results of the experiment indicated that the proposed approach can effectively improve the DTI prediction accuracy and can be used as a practical tool for the research and design of new drugs.

Investigating Monoliths (Vinyl Azlactone-co-Ethylene Dimethacrylate) as a Support for Enzymes and Drugs, for Proteomics and Drug-Target Studies

Prior to mass spectrometry, on-line sample preparation can be beneficial to reduce manual steps, increase speed, and enable analysis of limited sample amounts. For example, bottom-up proteomics sample preparation and analysis can be accelerated by digesting proteins to peptides in an on-line enzyme reactor. We here focus on low-backpressure 100 μm inner diameter (ID) × 160 mm, 180 μm ID × 110 mm or 250 μm ID × 140 mm vinyl azlactone-co-ethylene dimethacrylate [poly(VDM-co-EDMA)] monoliths as supports for immobilizing of additional molecules (i.e., proteases or drugs), as the monolith was expected to have few unspecific interactions. For on-line protein digestion, monolith supports immobilized with trypsin enzyme were found to be suited, featuring the expected characteristics of the material, i.e., low backpressure and low carry-over. Serving as a functionalized sample loop, the monolith units were very simple to connect on-line with liquid chromatography. However, for on-line target deconvolution, the monolithic support immobilized with a Wnt pathway inhibitor was associated with numerous secondary interactions when exploring the possibility of selectively trapping target proteins by drug-target interactions. Our initial observations suggest that (poly(VDM-co-EDMA)) monoliths are promising for e.g., on-line bottom-up proteomics, but not a "fit-for-all" material. We also discuss issues related to the repeatability of monolith-preparations.

Identification of sitagliptin binding proteins by affinity purification mass spectrometry

Type 2 diabetes mellitus (T2DM) is recognized as a serious public health concern with increasing incidence. The dipeptidyl peptidase-4 (DPP-4) inhibitor sitagliptin has been used for the treatment of T2DM worldwide. Although sitagliptin has excellent therapeutic outcome, adverse effects are observed. In addition, previous studies have suggested that sitagliptin may have pleiotropic effects other than treating T2DM. These pieces of evidence point to the importance of further investigation of the molecular mechanisms of sitagliptin, starting from the identification of sitagliptin-binding proteins. In this study, by combining affinity purification mass spectrometry (AP-MS) and stable isotope labeling by amino acids in cell culture (SILAC), we discover seven high-confidence targets that can interact with sitagliptin. Surface plasmon resonance (SPR) assay confirms the binding of sitagliptin to three proteins, i. e., LYPLAL1, TCP1, and CCAR2, with binding affinities (K _D) ranging from 50.1 μM to 1490 μM. Molecular docking followed by molecular dynamic (MD) simulation reveals hydrogen binding between sitagliptin and the catalytic triad of LYPLAL1, and also between sitagliptin and the P-loop of ATP-binding pocket of TCP1. Molecular mechanics Poisson-Boltzmann Surface Area (MMPBSA) analysis indicates that sitagliptin can stably bind to LYPLAL1 and TCP1 in active sites, which may have an impact on the functions of these proteins. SPR analysis validates the binding affinity of sitagliptin to TCP1 mutant D88A is ~10 times lower than that to the wild-type TCP1. Our findings provide insights into the sitagliptin-targets interplay and demonstrate the potential of sitagliptin in regulating gluconeogenesis and in anti-tumor drug development.

Virtual and In Vitro Screening of Natural Products Identifies Indole and Benzene Derivatives as Inhibitors of SARS-CoV-2 Main Protease (Mpro)

The rapid spread of the coronavirus disease 2019 (COVID-19) resulted in serious health, social, and economic consequences. While the development of effective vaccines substantially reduced the severity of symptoms and the associated deaths, we still urgently need effective drugs to further reduce the number of casualties associated with SARS-CoV-2 infections. Machine learning methods both improved and sped up all the different stages of the drug discovery processes by performing complex analyses with enormous datasets. Natural products (NPs) have been used for treating diseases and infections for thousands of years and represent a valuable resource for drug discovery when combined with the current computation advancements. Here, a dataset of 406,747 unique NPs was screened against the SARS-CoV-2 main protease (M^pro) crystal structure (6lu7) using a combination of ligand- and structural-based virtual screening. Based on 1) the predicted binding affinities of the NPs to the M^pro, 2) the types and number of interactions with the M^pro amino acids that are critical for its function, and 3) the desirable pharmacokinetic properties of the NPs, we identified the top 20 candidates that could potentially inhibit the M^pro protease function. A total of 7 of the 20 top candidates were subjected to in vitro protease inhibition assay and 4 of them (4/7; 57%), including two beta carbolines, one N-alkyl indole, and one Benzoic acid ester, had significant inhibitory activity against M^pro protease. These four NPs could be developed further for the treatment of COVID-19 symptoms.

Identifying potential drug-target interactions based on ensemble deep learning

Introduction

Drug-target interaction prediction is one important step in drug research and development. Experimental methods are time consuming and laborious.

Methods

In this study, we developed a novel DTI prediction method called EnGDD by combining initial feature acquisition, dimensional reduction, and DTI classification based on Gradient boosting neural network, Deep neural network, and Deep Forest.

Results

EnGDD was compared with seven stat-of-the-art DTI prediction methods (BLM-NII, NRLMF, WNNGIP, NEDTP, DTi2Vec, RoFDT, and MolTrans) on the nuclear receptor, GPCR, ion channel, and enzyme datasets under cross validations on drugs, targets, and drug-target pairs, respectively. EnGDD computed the best recall, accuracy, F1-score, AUC, and AUPR under the majority of conditions, demonstrating its powerful DTI identification performance. EnGDD predicted that D00182 and hsa2099, D07871 and hsa1813, DB00599 and hsa2562, D00002 and hsa10935 have a higher interaction probabilities among unknown drug-target pairs and may be potential DTIs on the four datasets, respectively. In particular, D00002 (Nadide) was identified to interact with hsa10935 (Mitochondrial peroxiredoxin3) whose up-regulation might be used to treat neurodegenerative diseases. Finally, EnGDD was used to find possible drug targets for Parkinson's disease and Alzheimer's disease after confirming its DTI identification performance. The results show that D01277, D04641, and D08969 may be applied to the treatment of Parkinson's disease through targeting hsa1813 (dopamine receptor D2) and D02173, D02558, and D03822 may be the clues of treatment for patients with Alzheimer's disease through targeting hsa5743 (prostaglandinendoperoxide synthase 2). The above prediction results need further biomedical validation.

Discussion

We anticipate that our proposed EnGDD model can help discover potential therapeutic clues for various diseases including neurodegenerative diseases.

Towards a more general drug target interaction prediction model using transfer learning

The topic of Drug-Target Interaction (DTI) topic has emerged nowadays since the COVID-19 outbreaks. DTI is one of the stages of finding a new cure for a recent disease. It determines whether a chemical compound would affect a particular protein, known as binding affinity. Recently, significant efforts have been devoted to artificial intelligence (AI) powered DTI. However, the use of transfer learning in DTI has not been explored extensively. This paper aims to make a more general DTI model by investigating DTI prediction method using Transfer learning. Three popular models will be tested and observed: CNN, RNN, and Transformer. Those models combined in several scenarios involving two extensive public datasets on DTI (BindingDB and DAVIS) to find the most optimum architecture. In our finding, combining the CNN model and BindingDB as the source data became the most recommended pre-trained model for real DTI cases. This conclusion was proved with the 6% AUPRC increase after fine-tuning the BindingDB pre-trained model to DAVIS dataset than without pre-training the model first.

GraphATT-DTA: Attention-Based Novel Representation of Interaction to Predict Drug-Target Binding Affinity

Drug-target binding affinity (DTA) prediction is an essential step in drug discovery. Drug-target protein binding occurs at specific regions between the protein and drug, rather than the entire protein and drug. However, existing deep-learning DTA prediction methods do not consider the interactions between drug substructures and protein sub-sequences. This work proposes GraphATT-DTA, a DTA prediction model that constructs the essential regions for determining interaction affinity between compounds and proteins, modeled with an attention mechanism for interpretability. We make the model consider the local-to-global interactions with the attention mechanism between compound and protein. As a result, GraphATT-DTA shows an improved prediction of DTA performance and interpretability compared with state-of-the-art models. The model is trained and evaluated with the Davis dataset, the human kinase dataset; an external evaluation is achieved with the independently proposed human kinase dataset from the BindingDB dataset.