Building Machine-Learning Scoring Functions for Structure-Based Prediction of Intermolecular Binding Affinity

SAnDReS 2.0: Development of machine-learning models to explore the scoring function space

Classical scoring functions may exhibit low accuracy in determining ligand binding affinity for proteins. The availability of both protein-ligand structures and affinity data make it possible to develop machine-learning models focused on specific protein systems with superior predictive performance. Here, we report a new methodology named SAnDReS that combines AutoDock Vina 1.2 with 54 regression methods available in Scikit-Learn to calculate binding affinity based on protein-ligand structures. This approach allows exploration of the scoring function space. SAnDReS generates machine-learning models based on crystal, docked, and AlphaFold-generated structures. As a proof of concept, we examine the performance of SAnDReS-generated models in three case studies. For all three cases, our models outperformed classical scoring functions. Also, SAnDReS-generated models showed predictive performance close to or better than other machine-learning models such as KDEEP, CSM-lig, and ΔVinaRF20. SAnDReS 2.0 is available to download at https://github.com/azevedolab/sandres.

Exploring Scoring Function Space: Developing Computational Models for Drug Discovery

BACKGROUND

The idea of scoring function space established a systems-level approach to address the development of models to predict the affinity of drug molecules by those interested in drug discovery.

OBJECTIVE

Our goal here is to review the concept of scoring function space and how to explore it to develop machine learning models to address protein-ligand binding affinity.

METHODS

We searched the articles available in PubMed related to the scoring function space. We also utilized crystallographic structures found in the protein data bank (PDB) to represent the protein space.

RESULTS

The application of systems-level approaches to address receptor-drug interactions allows us to have a holistic view of the process of drug discovery. The scoring function space adds flexibility to the process since it makes it possible to see drug discovery as a relationship involving mathematical spaces.

CONCLUSION

The application of the concept of scoring function space has provided us with an integrated view of drug discovery methods. This concept is useful during drug discovery, where we see the process as a computational search of the scoring function space to find an adequate model to predict receptor-drug binding affinity.

Mpropred: A machine learning (ML) driven Web-App for bioactivity prediction of SARS-CoV-2 main protease (Mpro) antagonists

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic emerged in 2019 and still requiring treatments with fast clinical translatability. Frequent occurrence of mutations in spike glycoprotein of SARS-CoV-2 led the consideration of an alternative therapeutic target to combat the ongoing pandemic. The main protease (Mpro) is such an attractive drug target due to its importance in maturating several polyproteins during the replication process. In the present study, we used a classification structure-activity relationship (CSAR) model to find substructures that leads to to anti-Mpro activities among 758 non-redundant compounds. A set of 12 fingerprints were used to describe Mpro inhibitors, and the random forest approach was used to build prediction models from 100 distinct data splits. The data set's modelability (MODI index) was found to be robust, with a value of 0.79 above the 0.65 threshold. The accuracy (89%), sensitivity (89%), specificity (73%), and Matthews correlation coefficient (79%) used to calculate the prediction performance, was also found to be statistically robust. An extensive analysis of the top significant descriptors unveiled the significance of methyl side chains, aromatic ring and halogen groups for Mpro inhibition. Finally, the predictive model is made publicly accessible as a web-app named Mpropred in order to allow users to predict the bioactivity of compounds against SARS-CoV-2 Mpro. Later, CMNPD, a marine compound database was screened by our app to predict bioactivity of all the compounds and results revealed significant correlation with their binding affinity to Mpro. Molecular dynamics (MD) simulation and molecular mechanics/Poisson Boltzmann surface area (MM/PBSA) analysis showed improved properties of the complexes. Thus, the knowledge and web-app shown herein can be used to develop more effective and specific inhibitors against the SARS-CoV-2 Mpro. The web-app can be accessed from https://share.streamlit.io/nadimfrds/mpropred/Mpropred_app.py.

Designing drugs when there is low data availability: one-shot learning and other approaches to face the issues of a long-term concern

INTRODUCTION

Modern drug discovery generally is accessed by useful information from previous large databases or uncovering novel data. The lack of biological and/or chemical data tends to slow the development of scientific research and innovation. Here, approaches that may help provide solutions to generate or obtain enough relevant data or improve/accelerate existing methods within the last five years were reviewed.

AREAS COVERED

One-shot learning (OSL) approaches, structural modeling, molecular docking, scoring function space (SFS), molecular dynamics (MD), and quantum mechanics (QM) may be used to amplify the amount of available data to drug design and discovery campaigns, presenting methods, their perspectives, and discussions to be employed in the near future.

EXPERT OPINION

Recent works have successfully used these techniques to solve a range of issues in the face of data scarcity, including complex problems such as the challenging scenario of drug design aimed at intrinsically disordered proteins and the evaluation of potential adverse effects in a clinical scenario. These examples show that it is possible to improve and kickstart research from scarce available data to design and discover new potential drugs.

Can artificial intelligence accelerate preclinical drug discovery and precision medicine?

Precision medicine leverages molecular biomarkers for selecting optimal treatment strategies. Accordingly, stratifying patients into responder and non-responder has strongly accelerated drug discovery and drug approvals in the last two decades. Recently, the applications of artificial intelligence (AI) in healthcare have been promoting these processes, and continue to improve patient care through systematically analysing large-scale molecular data and electronic health records. In particular, preclinical pharmacogenomics data empowered AI to unfold its full potential.Areas covered: Here, we discuss the opportunities of AI in pharmacogenomics, drug discovery and precision medicine. In particular, we shed some light on the advancements in computational biomedicine from statistical, machine learning (ML) to complex deep learning (DL) models.Expert opinion: AI has already strongly impacted drug discovery, and will continue to revolutionise academic research and the pharmaceutical industry. Its algorithms aid the identification of novel treatment options through molecular signatures and thus pave the way for the next generation of precision medicine.

A brief review of protein-ligand interaction prediction

The task of identifying protein–ligand interactions (PLIs) plays a prominent role in the field of drug discovery. However, it is infeasible to identify potential PLIs via costly and laborious in vitro experiments. There is a need to develop PLI computational prediction approaches to speed up the drug discovery process. In this review, we summarize a brief introduction to various computation-based PLIs. We discuss these approaches, in particular, machine learning-based methods, with illustrations of different emphases based on mainstream trends. Moreover, we analyzed three research dynamics that can be further explored in future studies.

Identification of novel off targets of baricitinib and tofacitinib by machine learning with a focus on thrombosis and viral infection

As there are no clear on-target mechanisms that explain the increased risk for thrombosis and viral infection or reactivation associated with JAK inhibitors, the observed elevated risk may be a result of an off-target effect. Computational approaches combined with in vitro studies can be used to predict and validate the potential for an approved drug to interact with additional (often unwanted) targets and identify potential safety-related concerns. Potential off-targets of the JAK inhibitors baricitinib and tofacitinib were identified using two established machine learning approaches based on ligand similarity. The identified targets related to thrombosis or viral infection/reactivation were subsequently validated using in vitro assays. Inhibitory activity was identified for four drug-target pairs (PDE10A [baricitinib], TRPM6 [tofacitinib], PKN2 [baricitinib, tofacitinib]). Previously unknown off-target interactions of the two JAK inhibitors were identified. As the proposed pharmacological effects of these interactions include attenuation of pulmonary vascular remodeling, modulation of HCV response, and hypomagnesemia, the newly identified off-target interactions cannot explain an increased risk of thrombosis or viral infection/reactivation. While further evidence is required to explain both the elevated thrombosis and viral infection/reactivation risk, our results add to the evidence that these JAK inhibitors are promiscuous binders and highlight the potential for repurposing.

Deep Learning in Drug Design: Protein-Ligand Binding Affinity Prediction

Computational drug design relies on the calculation of binding strength between two biological counterparts especially a chemical compound, i.e., a ligand, and a protein. Predicting the affinity of protein-ligand binding with reasonable accuracy is crucial for drug discovery, and enables the optimization of compounds to achieve better interaction with their target protein. In this paper, we propose a data-driven framework named DeepAtom to accurately predict the protein-ligand binding affinity. With 3D Convolutional Neural Network (3D-CNN) architecture, DeepAtom could automatically extract binding related atomic interaction patterns from the voxelized complex structure. Compared with the other CNN based approaches, our light-weight model design effectively improves the model representational capacity, even with the limited available training data. We carried out validation experiments on the PDBbind v.2016 benchmark and the independent Astex Diverse Set. We demonstrate that the less feature engineering dependent DeepAtom approach consistently outperforms the other baseline scoring methods. We also compile and propose a new benchmark dataset to further improve the model performances. With the new dataset as training input, DeepAtom achieves Pearson's R=0.83 and RMSE=1.23 pK units on the PDBbind v.2016 core set. The promising results demonstrate that DeepAtom models can be potentially adopted in computational drug development protocols such as molecular docking and virtual screening.

ABCpred: a webserver for the discovery of acetyl- and butyryl-cholinesterase inhibitors

Alzheimer's disease (AD) is one of the most common forms of dementia and is associated with a decline in cognitive function and language ability. The deficiency of the cholinergic neurotransmitter known as acetylcholine (ACh) is associated with AD. Acetylcholinesterase (AChE) hydrolyses ACh and inhibits the cholinergic transmission. Furthermore, both AChE and butyrylcholinesterase (BChE) plays important roles in early and late stages of AD. Therefore, the inhibition of either or both cholinesterase enzymes represent a promising therapeutic route for treating AD. In this study, a large-scale classification structure-activity relationship model was developed to predict cholinesterase inhibitory activities as well as revealing important substructures governing their activities. Herein, a non-redundant dataset constituting 985 and 1056 compounds for AChE and BChE, respectively, was obtained from the ChEMBL database. These inhibitors were described by 12 sets of molecular fingerprints and predictive models were developed using the random forest algorithm. Evaluation of the model performance by means of Matthews correlation coefficient and consideration of the model's interpretability indicated that the SubstructureCount fingerprint was the most robust with five-fold cross-validated MCC of [0.76, 0.82] for AChE and BChE, respectively, and test MCC of [0.73, 0.97]. Feature interpretation revealed that the aromatic ring system, heterocyclic nitrogen containing compounds and amines are important for cholinesterase inhibition. Finally, the model was deployed as a publicly available webserver called the ABCpred at http://codes.bio/abcpred/ .

Computational Prediction of Binding Affinity for CDK2-ligand Complexes. A Protein Target for Cancer Drug Discovery

BACKGROUND

CDK2 participates in the control of eukaryotic cell-cycle progression. Due to the great interest in CDK2 for drug development and the relative easiness in crystallizing this enzyme, we have over 400 structural studies focused on this protein target. This structural data is the basis for the development of computational models to estimate CDK2-ligand binding affinity.

OBJECTIVE

This work focuses on the recent developments in the application of supervised machine learning modeling to develop scoring functions to predict the binding affinity of CDK2.

METHOD

We employed the structures available at the protein data bank and the ligand information accessed from the BindingDB, Binding MOAD, and PDBbind to evaluate the predictive performance of machine learning techniques combined with physical modeling used to calculate binding affinity. We compared this hybrid methodology with classical scoring functions available in docking programs.

RESULTS

Our comparative analysis of previously published models indicated that a model created using a combination of a mass-spring system and cross-validated Elastic Net to predict the binding affinity of CDK2-inhibitor complexes outperformed classical scoring functions available in AutoDock4 and AutoDock Vina.

CONCLUSION

All studies reviewed here suggest that targeted machine learning models are superior to classical scoring functions to calculate binding affinities. Specifically for CDK2, we see that the combination of physical modeling with supervised machine learning techniques exhibits improved predictive performance to calculate the protein-ligand binding affinity. These results find theoretical support in the application of the concept of scoring function space.

Machine Learning-Based Scoring Functions, Development and Applications with SAnDReS

BACKGROUND

Analysis of atomic coordinates of protein-ligand complexes can provide three-dimensional data to generate computational models to evaluate binding affinity and thermodynamic state functions. Application of machine learning techniques can create models to assess protein-ligand potential energy and binding affinity. These methods show superior predictive performance when compared with classical scoring functions available in docking programs.

OBJECTIVE

Our purpose here is to review the development and application of the program SAnDReS. We describe the creation of machine learning models to assess the binding affinity of protein-ligand complexes.

METHODS

SAnDReS implements machine learning methods available in the scikit-learn library. This program is available for download at https://github.com/azevedolab/sandres. SAnDReS uses crystallographic structures, binding and thermodynamic data to create targeted scoring functions.

RESULTS

Recent applications of the program SAnDReS to drug targets such as Coagulation factor Xa, cyclin-dependent kinases and HIV-1 protease were able to create targeted scoring functions to predict inhibition of these proteins. These targeted models outperform classical scoring functions.

CONCLUSION

Here, we reviewed the development of machine learning scoring functions to predict binding affinity through the application of the program SAnDReS. Our studies show the superior predictive performance of the SAnDReS-developed models when compared with classical scoring functions available in the programs such as AutoDock4, Molegro Virtual Docker and AutoDock Vina.

Machine learning classification can reduce false positives in structure-based virtual screening

With the recent explosion in the size of libraries available for screening, virtual screening is positioned to assume a more prominent role in early drug discovery's search for active chemical matter. In typical virtual screens, however, only about 12% of the top-scoring compounds actually show activity when tested in biochemical assays. We argue that most scoring functions used for this task have been developed with insufficient thoughtfulness into the datasets on which they are trained and tested, leading to overly simplistic models and/or overtraining. These problems are compounded in the literature because studies reporting new scoring methods have not validated their models prospectively within the same study. Here, we report a strategy for building a training dataset (D-COID) that aims to generate highly compelling decoy complexes that are individually matched to available active complexes. Using this dataset, we train a general-purpose classifier for virtual screening (vScreenML) that is built on the XGBoost framework. In retrospective benchmarks, our classifier shows outstanding performance relative to other scoring functions. In a prospective context, nearly all candidate inhibitors from a screen against acetylcholinesterase show detectable activity; beyond this, 10 of 23 compounds have IC₅₀ better than 50 μM. Without any medicinal chemistry optimization, the most potent hit has IC₅₀ 280 nM, corresponding to K_i of 173 nM. These results support using the D-COID strategy for training classifiers in other computational biology tasks, and for vScreenML in virtual screening campaigns against other protein targets. Both D-COID and vScreenML are freely distributed to facilitate such efforts.

The impact of compound library size on the performance of scoring functions for structure-based virtual screening

Larger training datasets have been shown to improve the accuracy of machine learning (ML)-based scoring functions (SFs) for structure-based virtual screening (SBVS). In addition, massive test sets for SBVS, known as ultra-large compound libraries, have been demonstrated to enable the fast discovery of selective drug leads with low-nanomolar potency. This proof-of-concept was carried out on two targets using a single docking tool along with its SF. It is thus unclear whether this high level of performance would generalise to other targets, docking tools and SFs. We found that screening a larger compound library results in more potent actives being identified in all six additional targets using a different docking tool along with its classical SF. Furthermore, we established that a way to improve the potency of the retrieved molecules further is to rank them with more accurate ML-based SFs (we found this to be true in four of the six targets; the difference was not significant in the remaining two targets). A 3-fold increase in average hit rate across targets was also achieved by the ML-based SFs. Lastly, we observed that classical and ML-based SFs often find different actives, which supports using both types of SFs on those targets.

HCVpred: A web server for predicting the bioactivity of hepatitis C virus NS5B inhibitors

Hepatitis C virus (HCV) is one of the major causes of liver disease affecting an estimated 170 million people culminating in 300,000 deaths from cirrhosis or liver cancer. NS5B is one of three potential therapeutic targets against HCV (i.e., the other two being NS3/4A and NS5A) that is central to viral replication. In this study, we developed a classification structure-activity relationship (CSAR) model for identifying substructures giving rise to anti-HCV activities among a set of 578 non-redundant compounds. NS5B inhibitors were described by a set of 12 fingerprint descriptors and predictive models were constructed from 100 independent data splits using the random forest algorithm. The modelability (MODI index) of the data set was determined to be robust with a value of 0.88 exceeding established threshold of 0.65. The predictive performance was deduced by the accuracy, sensitivity, specificity, and Matthews correlation coefficient, which was found to be statistically robust (i.e., the former three parameters afforded values in excess of 0.8 while the latter statistical parameter provided a value >0.7). An in-depth analysis of the top 20 important descriptors revealed that aromatic ring and alkyl side chains are important for NS5B inhibition. Finally, the predictive model is deployed as a publicly accessible HCVpred web server (available at http://codes.bio/hcvpred/) that would allow users to predict the biological activity as being active or inactive against HCV NS5B. Thus, the knowledge and web server presented herein can be used in the design of more potent and specific drugs against the HCV NS5B.