Bisindolylmaleimide IX: A novel anti-SARS-CoV2 agent targeting viral main protease 3CLpro demonstrated by virtual screening pipeline and in-vitro validation assays.

SARS-CoV-2, the virus that causes COVID-19 consists of several enzymes with essential functions within its proteome. Here, we focused on repurposing approved and investigational drugs/compounds. We targeted seven proteins with enzymatic activities known to be essential at different stages of the viral cycle including PLpro, 3CLpro, RdRP, Helicase, ExoN, NendoU, and 2'-O-MT. For virtual screening, energy minimization of a crystal structure of the modeled protein was carried out using the Protein Preparation Wizard (Schrodinger LLC 2020-1). Following active site selection based on data mining and COACH predictions, we performed a high-throughput virtual screen of drugs and investigational molecules (n = 5903). The screening was performed against viral targets using three sequential docking modes (i.e., HTVS, SP, and XP). Virtual screening identified ∼290 potential inhibitors based on the criteria of energy, docking parameters, ligand, and binding site strain and score. Drugs specific to each target protein were further analyzed for binding free energy perturbation by molecular mechanics (prime MM-GBSA) and pruning the hits to the top 32 candidates. The top lead from each target pool was further subjected to molecular dynamics simulation using the Desmond module. The resulting top eight hits were tested for their SARS-CoV-2 anti-viral activity in-vitro. Among these, a known inhibitor of protein kinase C isoforms, Bisindolylmaleimide IX (BIM IX), was found to be a potent inhibitor of SARS-CoV-2. Further, target validation through enzymatic assays confirmed 3CLpro to be the target. This is the first study that has showcased BIM IX as a COVID-19 inhibitor thereby validating our pipeline.

Performance of Force-Field- and Machine Learning-Based Scoring Functions in Ranking MAO-B Protein-Inhibitor Complexes in Relevance to Developing Parkinson's Therapeutics.

Monoamine oxidase B (MAOB) is expressed in the mitochondrial membrane and has a key role in degrading various neurologically active amines such as benzylamine, phenethylamine and dopamine with the help of Flavin adenine dinucleotide (FAD) cofactor. The Parkinson's disease associated symptoms can be treated using inhibitors of MAO-B as the dopamine degradation can be reduced. Currently, many inhibitors are available having micromolar to nanomolar binding affinities. However, still there is demand for compounds with superior binding affinity and binding specificity with favorable pharmacokinetic properties for treating Parkinson's disease and computational screening methods can be majorly recruited for this. However, the accuracy of currently available force-field methods for ranking the inhibitors or lead drug-like compounds should be improved and novel methods for screening compounds need to be developed. We studied the performance of various force-field-based methods and data driven approaches in ranking about 3753 compounds having activity against the MAO-B target. The binding affinities computed using autodock and autodock-vina are shown to be non-reliable. The force-field-based MM-GBSA also under-performs. However, certain machine learning approaches, in particular KNN, are found to be superior, and we propose KNN as the most reliable approach for ranking the complexes to reasonable accuracy. Furthermore, all the employed machine learning approaches are also computationally less demanding.

Dockground Tool for Development and Benchmarking of Protein Docking Procedures.

Databases of protein-protein complexes are essential for the development of protein modeling/docking techniques. Such databases provide a knowledge base for docking algorithms, intermolecular potentials, search procedures, scoring functions, and refinement protocols. Development of docking techniques requires systematic validation of the modeling protocols on carefully curated benchmark sets of complexes. We present a description and a guide to the DOCKGROUND resource ( http://dockground.compbio.ku.edu ) for structural modeling of protein interactions. The resource integrates various datasets of protein complexes and other data for the development and testing of protein docking techniques. The sets include bound complexes, experimentally determined unbound, simulated unbound, model-model complexes, and docking decoys. The datasets are available to the user community through a Web interface.

Accelerated CDOCKER with GPUs, Parallel Simulated Annealing, and Fast Fourier Transforms.

Fast Fourier transform (FFT)-based protein ligand docking together with parallel simulated annealing for both rigid and flexible receptor docking are implemented on graphical processing unit (GPU) accelerated platforms to significantly enhance the throughput of the CDOCKER and flexible CDOCKER - the docking algorithms in the CHARMM program for biomolecule modeling. The FFT-based approach for docking, first applied in protein-protein docking to efficiently search for the binding position and orientation of proteins, is adapted here to search ligand translational and rotational spaces given a ligand conformation in protein-ligand docking. Running on GPUs, our implementation of FFT docking in CDOCKER achieves a 15 000 fold speedup in the ligand translational and rotational space search in protein-ligand docking problems. With this significant speedup it becomes practical to exhaustively search ligand translational and rotational space when docking a rigid ligand into a protein receptor. We demonstrate in this paper that this provides an efficient way to calculate an upper bound for docking accuracy in the assessment of scoring functions for protein-ligand docking, which can be useful for improving scoring functions. The parallel molecular dynamics (MD) simulated annealing, also running on GPUs, aims to accelerate the search algorithm in CDOCKER by running MD simulated annealing in parallel on GPUs. When utilized as part of the general CDOCKER docking protocol, acceleration in excess of 20 times is achieved. With this acceleration, we demonstrate that the performance of CDOCKER for redocking is significantly improved compared with three other popular protein-ligand docking programs on two widely used protein ligand complex data sets: the Astex diverse set and the SB2012 test set. The flexible CDOCKER is similarly improved by the parallel MD simulated annealing on GPUs. Based on the results presented here, we suggest that the accelerated CDOCKER platform provides a highly competitive docking engine for both rigid-receptor and flexible-receptor docking studies and will further facilitate continued improvement in the physics-based scoring function employed in CDOCKER docking studies.

Comprehensive Evaluation of Fourteen Docking Programs on Protein-Peptide Complexes.

A large number of protein-protein interactions (PPIs) are mediated by the interactions between proteins and peptide segments binding partners, and therefore determination of protein-peptide interactions (PpIs) is quite crucial to elucidate important biological processes and design peptides or peptidomimetic drugs that can modulate PPIs. Nowadays, as a powerful computation tool, molecular docking has been widely utilized to predict the binding structures of protein-peptide complexes. However, although a number of docking programs have been available, the systematic study on the assessment of their performance for PpIs has never been reported. In this study, a benchmark data set called PepSet consisting of 185 protein-peptide complexes with peptide length ranging from 5 to 20 residues was employed to evaluate the performance of 14 docking programs, including three protein-protein docking programs (ZDOCK, FRODOCK, and HawkDock), three small molecule docking programs (GOLD, Surflex-Dock, and AutoDock Vina), and eight protein-peptide docking programs (GalaxyPepDock, MDockPeP, HPEPDOCK, CABS-dock, pepATTRACT, DINC, AutoDock CrankPep (ADCP), and HADDOCK peptide docking). A new evaluation parameter, named IL_RMSD, was proposed to measure the docking accuracy with fnat (the fraction of native contacts). In global docking, HPEPDOCK performs the best for the entire data set and yields the success rates of 4.3%, 24.3%, and 55.7% at the top 1, 10, and 100 levels, respectively. In local docking, overall, ADCP achieves the best predictions and reaches the success rates of 11.9%, 37.3%, and 70.3% at the top 1, 10, and 100 levels, respectively. It is expected that our work can provide some helpful insights into the selection and development of improved docking programs for PpIs. The benchmark data set is freely available at http://cadd.zju.edu.cn/pepset/.

An open-source drug discovery platform enables ultra-large virtual screens.

On average, an approved drug currently costs US$2-3 billion and takes more than 10 years to develop1. In part, this is due to expensive and time-consuming wet-laboratory experiments, poor initial hit compounds and the high attrition rates in the (pre-)clinical phases. Structure-based virtual screening has the potential to mitigate these problems. With structure-based virtual screening, the quality of the hits improves with the number of compounds screened2. However, despite the fact that large databases of compounds exist, the ability to carry out large-scale structure-based virtual screening on computer clusters in an accessible, efficient and flexible manner has remained difficult. Here we describe VirtualFlow, a highly automated and versatile open-source platform with perfect scaling behaviour that is able to prepare and efficiently screen ultra-large libraries of compounds. VirtualFlow is able to use a variety of the most powerful docking programs. Using VirtualFlow, we prepared one of the largest and freely available ready-to-dock ligand libraries, with more than 1.4 billion commercially available molecules. To demonstrate the power of VirtualFlow, we screened more than 1 billion compounds and identified a set of structurally diverse molecules that bind to KEAP1 with submicromolar affinity. One of the lead inhibitors (iKeap1) engages KEAP1 with nanomolar affinity (dissociation constant (Kd) = 114 nM) and disrupts the interaction between KEAP1 and the transcription factor NRF2. This illustrates the potential of VirtualFlow to access vast regions of the chemical space and identify molecules that bind with high affinity to target proteins.

Improvement in predicting drug sensitivity changes associated with protein mutations using a molecular dynamics based alchemical mutation method.

While molecular-targeted drugs have demonstrated strong therapeutic efficacy against diverse diseases such as cancer and infection, the appearance of drug resistance associated with genetic variations in individual patients or pathogens has severely limited their clinical efficacy. Therefore, precision medicine approaches based on the personal genomic background provide promising strategies to enhance the effectiveness of molecular-targeted therapies. However, identifying drug resistance mutations in individuals by combining DNA sequencing and in vitro analyses is generally time consuming and costly. In contrast, in silico computation of protein-drug binding free energies allows for the rapid prediction of drug sensitivity changes associated with specific genetic mutations. Although conventional alchemical free energy computation methods have been used to quantify mutation-induced drug sensitivity changes in some protein targets, these methods are often adversely affected by free energy convergence. In this paper, we demonstrate significant improvements in prediction performance and free energy convergence by employing an alchemical mutation protocol, MutationFEP, which directly estimates binding free energy differences associated with protein mutations in three types of a protein and drug system. The superior performance of MutationFEP appears to be attributable to its more-moderate perturbation scheme. Therefore, this study provides a deeper level of insight into computer-assisted precision medicine.

P3DOCK: a protein-RNA docking webserver based on template-based and template-free docking.

MOTIVATION: The main function of protein-RNA interaction is to regulate the expression of genes. Therefore, studying protein-RNA interactions is of great significance. The information of three-dimensional (3D) structures reveals that atomic interactions are particularly important. The calculation method for modeling a 3D structure of a complex mainly includes two strategies: free docking and template-based docking. These two methods are complementary in protein-protein docking. Therefore, integrating these two methods may improve the prediction accuracy. RESULTS: In this article, we compare the difference between the free docking and the template-based algorithm. Then we show the complementarity of these two methods. Based on the analysis of the calculation results, the transition point is confirmed and used to integrate two docking algorithms to develop P3DOCK. P3DOCK holds the advantages of both algorithms. The results of the three docking benchmarks show that P3DOCK is better than those two non-hybrid docking algorithms. The success rate of P3DOCK is also higher (3-20%) than state-of-the-art hybrid and non-hybrid methods. Finally, the hierarchical clustering algorithm is utilized to cluster the P3DOCK's decoys. The clustering algorithm improves the success rate of P3DOCK. For ease of use, we provide a P3DOCK webserver, which can be accessed at www.rnabinding.com/P3DOCK/P3DOCK.html. An integrated protein-RNA docking benchmark can be downloaded from http://rnabinding.com/P3DOCK/benchmark.html. AVAILABILITY AND IMPLEMENTATION: www.rnabinding.com/P3DOCK/P3DOCK.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Model Optimization and In Silico Analysis of Potential Dipeptidyl Peptidase IV Antagonists from GC-MS Identified Compounds in Nauclea latifolia Leaf Extracts.

Dipeptidyl peptidase IV (DPP-IV) is a pharmacotherapeutic target in type 2 diabetes. Inhibitors of this enzyme constitute a new class of drugs used in the treatment and management of type 2 diabetes. In this study, phytocompounds in Nauclea latifolia (NL) leaf extracts, identified using gas chromatography-mass spectroscopy (GC-MS), were tested for potential antagonists of DPP-IV via in silico techniques. Phytocompounds present in N. latifolia aqueous (NLA) and ethanol (NLE) leaf extracts were identified using GC-MS. DPP-IV model optimization and molecular docking of the identified compounds/standard inhibitors in the binding pocket was simulated. Drug-likeness, pharmacokinetic and pharmacodynamic properties of promising docked leads were also predicted. Results showed the presence of 50 phytocompounds in NL extracts of which only 2-O-p-methylphenyl-1-thio-ß-d-glucoside, 3-tosylsedoheptulose, 4-benzyloxy-6-hydroxymethyl-tetrahydropyran-2,3,5-triol and vitamin E exhibited comparable or better binding iGEMDOCK and AutoDock Vina scores than the clinically prescribed standards. These four compounds exhibited promising drug-likeness as well as absorption, distribution, metabolism, excretion and toxicity (ADMET) properties suggesting their candidature as novel leads for developing DPP-IV inhibitors.

LC/QTOF/MS/MS characterization, molecular docking and in silico toxicity prediction studies on degradation products of anagliptin.

Pharmaceutical drugs are potential molecules with specific biological activity. However, long-term use of these chemical molecules can affect the human physiological system because of their increased levels in the human body. Therefore, identification and structure elucidation of impurities or degradation products should be taken into consideration in order to assure drug safety. The present study assessed the degradation behaviour of dipeptidyl peptidase-4 (DPP-4) inhibitor anagliptin under different stress conditions as per ICH guidelines Q1A (R2) followed by elucidation of the structure of degradation products. All the stress samples were analysed by using UPLC/PDA. The superior separation of drug from its degradation products was attained with time programmed gradient elution on BEH C18 (100 mm × 2.1 mm, 1.7 µm) column using 10 mM ammonium formate (aqueous) and acetonitrile (organic) as the mobile phase components. All the degradation products of anagliptin were characterized using LC/QTOF/MS/MS. In addition, the activity and toxicity of degradation products were determined through molecular docking and in silico toxicity prediction studies, respectively. The developed UPLC/PDA method was validated as per ICH guidelines in terms of specificity, accuracy, precision, linearity and robustness.

The Impact of Protein Structure and Sequence Similarity on the Accuracy of Machine-Learning Scoring Functions for Binding Affinity Prediction.

It has recently been claimed that the outstanding performance of machine-learning scoring functions (SFs) is exclusively due to the presence of training complexes with highly similar proteins to those in the test set. Here, we revisit this question using 24 similarity-based training sets, a widely used test set, and four SFs. Three of these SFs employ machine learning instead of the classical linear regression approach of the fourth SF (X-Score which has the best test set performance out of 16 classical SFs). We have found that random forest (RF)-based RF-Score-v3 outperforms X-Score even when 68% of the most similar proteins are removed from the training set. In addition, unlike X-Score, RF-Score-v3 is able to keep learning with an increasing training set size, becoming substantially more predictive than X-Score when the full 1105 complexes are used for training. These results show that machine-learning SFs owe a substantial part of their performance to training on complexes with dissimilar proteins to those in the test set, against what has been previously concluded using the same data. Given that a growing amount of structural and interaction data will be available from academic and industrial sources, this performance gap between machine-learning SFs and classical SFs is expected to enlarge in the future.

A Computational-Based Method for Predicting Drug-Target Interactions by Using Stacked Autoencoder Deep Neural Network.

Identifying the interaction between drugs and target proteins is an important area of drug research, which provides a broad prospect for low-risk and faster drug development. However, due to the limitations of traditional experiments when revealing drug-protein interactions (DTIs), the screening of targets not only takes a lot of time and money but also has high false-positive and false-negative rates. Therefore, it is imperative to develop effective automatic computational methods to accurately predict DTIs in the postgenome era. In this article, we propose a new computational method for predicting DTIs from drug molecular structure and protein sequence by using the stacked autoencoder of deep learning, which can adequately extract the raw data information. The proposed method has the advantage that it can automatically mine the hidden information from protein sequences and generate highly representative features through iterations of multiple layers. The feature descriptors are then constructed by combining the molecular substructure fingerprint information, and fed into the rotation forest for accurate prediction. The experimental results of fivefold cross-validation indicate that the proposed method achieves superior performance on gold standard data sets (enzymes, ion channels, GPCRs [G-protein-coupled receptors], and nuclear receptors) with accuracy of 0.9414, 0.9116, 0.8669, and 0.8056, respectively. We further comprehensively explore the performance of the proposed method by comparing it with other feature extraction algorithms, state-of-the-art classifiers, and other excellent methods on the same data set. The excellent comparison results demonstrate that the proposed method is highly competitive when predicting drug-target interactions.

Analysis of the quality of crystallographic data and the limitations of structural models.

Crystal structures provide visual models of biological macromolecules, which are widely used to interpret data from functional studies and generate new mechanistic hypotheses. Because the quality of the collected x-ray diffraction data directly affects the reliability of the structural model, it is essential that the limitations of the models are carefully taken into account when making interpretations. Here we use the available crystal structures of members of the glutamate transporter family to illustrate the importance of inspecting the data that underlie the structural models. Crystal structures of glutamate transporters in multiple different conformations have been solved, but most structures were determined at relatively low resolution, with deposited models based on crystallographic data of moderate quality. We use these examples to demonstrate the extent to which mechanistic interpretations can be made safely.

Applicability Domains and Consistent Structure Generation.

In de novo molecular design, quantitative structure-activity relationship (QSAR) and quantitative structure-property relationship (QSPR) models are important to estimate activity and property values, respectively, for virtual molecular structures. To operate QSAR and QSPR models appropriately, applicability domains (ADs) of the models must be defined because estimated values are unreliable for virtual molecular structures that are dissimilar to structures of training compounds. We describe several methods to construct AD models, and then introduce a structure generator to change molecular structures that exist outside ADs to conform to ADs.

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR.

Neural networks have generated valuable Quantitative Structure-Activity/Property Relationships (QSAR/QSPR) models for a wide variety of small molecules and materials properties. They have grown in sophistication and many of their initial problems have been overcome by modern mathematical techniques. QSAR studies have almost always used so-called "shallow" neural networks in which there is a single hidden layer between the input and output layers. Recently, a new and potentially paradigm-shifting type of neural network based on Deep Learning has appeared. Deep learning methods have generated impressive improvements in image and voice recognition, and are now being applied to QSAR and QSAR modelling. This paper describes the differences in approach between deep and shallow neural networks, compares their abilities to predict the properties of test sets for 15 large drug data sets (the kaggle set), discusses the results in terms of the Universal Approximation theorem for neural networks, and describes how DNN may ameliorate or remove troublesome "activity cliffs" in QSAR data sets.

CGBVS-DNN: Prediction of Compound-protein Interactions Based on Deep Learning.

Computational prediction of compound-protein interactions (CPIs) is of great importance for drug design as the first step in in-silico screening. We previously proposed chemical genomics-based virtual screening (CGBVS), which predicts CPIs by using a support vector machine (SVM). However, the CGBVS has problems when training using more than a million datasets of CPIs since SVMs require an exponential increase in the calculation time and computer memory. To solve this problem, we propose the CGBVS-DNN, in which we use deep neural networks, a kind of deep learning technique, instead of the SVM. Deep learning does not require learning all input data at once because the network can be trained with small mini-batches. Experimental results show that the CGBVS-DNN outperformed the original CGBVS with a quarter million CPIs. Results of cross-validation show that the accuracy of the CGBVS-DNN reaches up to 98.2 % (σ<0.01) with 4 million CPIs.

Quantitative Structure-activity Relationship (QSAR) Models for Docking Score Correction.

In order to improve docking score correction, we developed several structure-based quantitative structure activity relationship (QSAR) models by protein-drug docking simulations and applied these models to public affinity data. The prediction models used descriptor-based regression, and the compound descriptor was a set of docking scores against multiple (∼600) proteins including nontargets. The binding free energy that corresponded to the docking score was approximated by a weighted average of docking scores for multiple proteins, and we tried linear, weighted linear and polynomial regression models considering the compound similarities. In addition, we tried a combination of these regression models for individual data sets such as IC50 , Ki , and %inhibition values. The cross-validation results showed that the weighted linear model was more accurate than the simple linear regression model. Thus, the QSAR approaches based on the affinity data of public databases should improve docking scores.

Reliability analysis and optimization of the consensus docking approach for the development of virtual screening studies.

Ligand-protein docking is one of the most common techniques used in virtual screening campaigns. Despite the large number of docking software available, there is still the need of improving the efficacy of docking-based virtual screenings. To date, only very few studies evaluated the possibility of combining the results of different docking methods to achieve higher success rates in virtual screening studies (consensus docking). In order to better understand the range of applicability of this approach, we carried out an extensive enriched database analysis using the DUD dataset. The consensus docking protocol was then refined by applying modifications concerning the calculation of pose consensus and the combination of docking methods included in the procedure. The results obtained suggest that this approach performs as well as the best available methods found in literature, confirming the idea that this procedure can be profitably used for the identification of new hit compounds.

Is It Reliable to Use Common Molecular Docking Methods for Comparing the Binding Affinities of Enantiomer Pairs for Their Protein Target?

Molecular docking is a computational chemistry method which has become essential for the rational drug design process. In this context, it has had great impact as a successful tool for the study of ligand-receptor interaction modes, and for the exploration of large chemical datasets through virtual screening experiments. Despite their unquestionable merits, docking methods are not reliable for predicting binding energies due to the simple scoring functions they use. However, comparisons between two or three complexes using the predicted binding energies as a criterion are commonly found in the literature. In the present work we tested how wise is it to trust the docking energies when two complexes between a target protein and enantiomer pairs are compared. For this purpose, a ligand library composed by 141 enantiomeric pairs was used, including compounds with biological activities reported against seven protein targets. Docking results using the software Glide (considering extra precision (XP), standard precision (SP), and high-throughput virtual screening (HTVS) modes) and AutoDock Vina were compared with the reported biological activities using a classification scheme. Our test failed for all modes and targets, demonstrating that an accurate prediction when binding energies of enantiomers are compared using docking may be due to chance. We also compared pairs of compounds with different molecular weights and found the same results.

Correcting the impact of docking pose generation error on binding affinity prediction.

BACKGROUND: Pose generation error is usually quantified as the difference between the geometry of the pose generated by the docking software and that of the same molecule co-crystallised with the considered protein. Surprisingly, the impact of this error on binding affinity prediction is yet to be systematically analysed across diverse protein-ligand complexes. RESULTS: Against commonly-held views, we have found that pose generation error has generally a small impact on the accuracy of binding affinity prediction. This is also true for large pose generation errors and it is not only observed with machine-learning scoring functions, but also with classical scoring functions such as AutoDock Vina. Furthermore, we propose a procedure to correct a substantial part of this error which consists of calibrating the scoring functions with re-docked, rather than co-crystallised, poses. In this way, the relationship between Vina-generated protein-ligand poses and their binding affinities is directly learned. As a result, test set performance after this error-correcting procedure is much closer to that of predicting the binding affinity in the absence of pose generation error (i.e. on crystal structures). We evaluated several strategies, obtaining better results for those using a single docked pose per ligand than those using multiple docked poses per ligand. CONCLUSIONS: Binding affinity prediction is often carried out on the docked pose of a known binder rather than its co-crystallised pose. Our results suggest than pose generation error is in general far less damaging for binding affinity prediction than it is currently believed. Another contribution of our study is the proposal of a procedure that largely corrects for this error. The resulting machine-learning scoring function is freely available at http://istar.cse.cuhk.edu.hk/rf-score-4.tgz and http://ballester.marseille.inserm.fr/rf-score-4.tgz .