Early Drug Discovery Research Articles

Drug-induced phospholipidosis confounds drug repurposing for SARS-CoV-2.

Repurposing drugs as treatments for COVID-19, the disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has drawn much attention. Beginning with sigma receptor ligands and expanding to other drugs from screening in the field, we became concerned that phospholipidosis was a shared mechanism underlying the antiviral activity of many repurposed drugs. For all of the 23 cationic amphiphilic drugs we tested, including hydroxychloroquine, azithromycin, amiodarone, and four others already in clinical trials, phospholipidosis was monotonically correlated with antiviral efficacy. Conversely, drugs active against the same targets that did not induce phospholipidosis were not antiviral. Phospholipidosis depends on the physicochemical properties of drugs and does not reflect specific target-based activities—rather, it may be considered a toxic confound in early drug discovery. Early detection of phospholipidosis could eliminate these artifacts, enabling a focus on molecules with therapeutic potential.

Zebrafish: An Attractive Model to Study Staphylococcus aureus Infection and Its Use as a Drug Discovery Tool.

Non-mammalian in vivo disease models are particularly popular in early drug discovery. Zebrafish (Danio rerio) is an attractive vertebrate model, the success of which is driven by several advantages, such as the optical transparency of larvae, the small and completely sequenced genome, the small size of embryos and larvae enabling high-throughput screening, and low costs. In this review, we highlight zebrafish models of Staphyloccoccus aureus infection, which are used in drug discovery and for studying disease pathogenesis and virulence. Further, these infection models are discussed in the context of other relevant zebrafish models for pharmacological and toxicological studies as part of early drug profiling. In addition, we examine key differences to commonly applied models of S. aureus infection based on invertebrate organisms, and we compare their frequency of use in academic research covering the period of January 2011 to January 2021.

The application of artificial intelligence and data integration in COVID-19 studies: a scoping review.

ObjectiveTo summarize how artificial intelligence (AI) is being applied in COVID-19 research and determine whether these AI applications integrated heterogenous data from different sources for modeling.Materials and MethodsWe searched 2 major COVID-19 literature databases, the National Institutes of Health’s LitCovid and the World Health Organization’s COVID-19 database on March 9, 2021. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline, 2 reviewers independently reviewed all the articles in 2 rounds of screening.ResultsIn the 794 studies included in the final qualitative analysis, we identified 7 key COVID-19 research areas in which AI was applied, including disease forecasting, medical imaging-based diagnosis and prognosis, early detection and prognosis (non-imaging), drug repurposing and early drug discovery, social media data analysis, genomic, transcriptomic, and proteomic data analysis, and other COVID-19 research topics. We also found that there was a lack of heterogenous data integration in these AI applications.DiscussionRisk factors relevant to COVID-19 outcomes exist in heterogeneous data sources, including electronic health records, surveillance systems, sociodemographic datasets, and many more. However, most AI applications in COVID-19 research adopted a single-sourced approach that could omit important risk factors and thus lead to biased algorithms. Integrating heterogeneous data for modeling will help realize the full potential of AI algorithms, improve precision, and reduce bias.ConclusionThere is a lack of data integration in the AI applications in COVID-19 research and a need for a multilevel AI framework that supports the analysis of heterogeneous data from different sources.

Predicting Successes and Failures of Clinical Trials With Outer Product–Based Convolutional Neural Network

Despite several improvements in the drug development pipeline over the past decade, drug failures due to unexpected adverse effects have rapidly increased at all stages of clinical trials. To improve the success rate of clinical trials, it is necessary to identify potential loser drug candidates that may fail at clinical trials. Therefore, we need to develop reliable models for predicting the outcomes of clinical trials of drug candidates, which have the potential to guide the drug discovery process. In this study, we propose an outer product–based convolutional neural network (OPCNN) model which integrates effectively chemical features of drugs and target-based features. The validation results via 10-fold cross-validations on the dataset used for a data-driven approach PrOCTOR proved that our OPCNN model performs quite well in terms of accuracy, F1-score, Matthews correlation coefficient (MCC), precision, recall, area under the curve (AUC) of the receiver operating characteristic, and area under the precision–recall curve (AUPRC). In particular, the proposed OPCNN model showed the best performance in terms of MCC, which is widely used in biomedicine as a performance metric and is a more reliable statistical measure. Through 10-fold cross-validation experiments, the accuracy of the OPCNN model is as high as 0.9758, F1 score is as high as 0.9868, the MCC reaches 0.8451, the precision is as high as 0.9889, the recall is as high as 0.9893, the AUC is as high as 0.9824, and the AUPRC is as high as 0.9979. The results proved that our OPCNN model shows significantly good prediction performance on outcomes of clinical trials and it can be quite helpful in early drug discovery.

Rational Control of Molecular Properties Is Mandatory to Exploit the Potential of PROTACs as Oral Drugs.

To obtain new oral drugs in the beyond rule of five space, PROTACs among others, molecular properties should be optimized in early drug discovery. Degraders call for design strategies which focus on intramolecular interaction and chameleonicity. In parallel, tailored revalidation of permeability assessment and prediction methods becomes fundamental in this innovative chemical space.

Exploring protein hotspots by optimized fragment pharmacophores

Fragment-based drug design has introduced a bottom-up process for drug development, with improved sampling of chemical space and increased effectiveness in early drug discovery. Here, we combine the use of pharmacophores, the most general concept of representing drug-target interactions with the theory of protein hotspots, to develop a design protocol for fragment libraries. The SpotXplorer approach compiles small fragment libraries that maximize the coverage of experimentally confirmed binding pharmacophores at the most preferred hotspots. The efficiency of this approach is demonstrated with a pilot library of 96 fragment-sized compounds (SpotXplorer0) that is validated on popular target classes and emerging drug targets. Biochemical screening against a set of GPCRs and proteases retrieves compounds containing an average of 70% of known pharmacophores for these targets. More importantly, SpotXplorer0 screening identifies confirmed hits against recently established challenging targets such as the histone methyltransferase SETD2, the main protease (3CLPro) and the NSP3 macrodomain of SARS-CoV-2.

DeepFrag: An Open-Source Browser App for Deep-Learning Lead Optimization.

Lead optimization, a critical step in early stage drug discovery, involves making chemical modifications to a small-molecule ligand to improve properties such as binding affinity. We recently developed DeepFrag, a deep-learning model capable of recommending such modifications. Though a powerful hypothesis-generating tool, DeepFrag is currently implemented in Python and so requires a certain degree of computational expertise. To encourage broader adoption, we have created the DeepFrag browser app, which provides a user-friendly graphical user interface that runs the DeepFrag model in users’ web browsers. The browser app does not require users to upload their molecular structures to a third-party server, nor does it require the separate installation of any third-party software. We are hopeful that the app will be a useful tool for both researchers and students. It can be accessed free of charge, without registration, at http://durrantlab.com/deepfrag. The source code is also available at http://git.durrantlab.com/jdurrant/deepfrag-app, released under the terms of the open-source Apache License, Version 2.0.

Ubiquitination Regulators Discovered by Virtual Screening for the Treatment of Cancer.

The ubiquitin-proteasome system oversees cellular protein degradation in order to regulate various critical processes, such as cell cycle control and DNA repair. Ubiquitination can serve as a marker for mutation, chemical damage, transcriptional or translational errors, and heat-induced denaturation. However, aberrant ubiquitination and degradation of tumor suppressor proteins may result in the growth and metastasis of cancer. Hence, targeting the ubiquitination cascade reaction has become a potential strategy for treating malignant diseases. Meanwhile, computer-aided methods have become widely accepted as fast and efficient techniques for early stage drug discovery. This review summarizes ubiquitination regulators that have been discovered via virtual screening and their applications for cancer treatment.

Comparative Proteomics Analysis of the Post‐Mitochondrial Fraction (S9 Fraction) of Human Lens‐Free Whole Eye and Liver

Background The increasing incidence of ocular diseases has accelerated research into therapeutic interventions for the eye. The knowledge of key metabolic mechanisms in the eye is vital for the successful ophthalmic drug discovery. The lens-free whole-eye post-mitochondrial fraction(S9 fraction) is one of the most widely used in vitro system for ocular metabolism studies during early drug discovery stage. As of now, no report of a comprehensive proteomics investigation exists for the pooled ocular S9 fraction. The objective of this study was to explore the human ocular S9 proteome and compare it with the human liver S9 proteome. Methods The global proteomics study was performed on the pooled ocular (n=11) and hepatic (n=50) S9 fraction samples in data-dependent acquisition mode using an optimized protocol. The samples were acquired in triplicate to minimize the technical variability. The acquired proteomics data were analyzed using Maxquant with 1% false discovery rate. A FASTA file containing whole human proteome, was used as the reference sequence for the search in the Andromeda database search tool. Additionally, another FASTA file containing 1187 proteins involved in the metabolism of endobiotics and xenobiotics was also searched to specifically identify the metabolic enzymes in both organs. The proteins were quantified using a generic label free quantification method (MaxLFQ) in Maxquant software. The metabolic enzymes common to both eye and liver S9 fractions were compared to determine the relative abundance values based on the resulting MaxLFQ intensities. The cut off for the differentially expressed proteins between the eye or liver was either <0.5 or >2. Results and Conclusion A total of 269 proteins (including 23 metabolic enzymes) were detected exclusively in the pooled eye S9, against 648 proteins in the liver S9 (including 174 metabolic enzymes), whereas 424 proteins (including 94 metabolic enzymes) were detected in both the organs. Among the common metabolic enzymes, 18 were predominant in the eye (Figure 1), 43 were predominant in the liver, while rest 33 were comparable in both samples. The major DMEs, viz., Cytochrome P450s, UDP-glucuronosyltransferases, and sulfotransferases, were not detected in the eye S9 fraction, instead isoforms of alcohol dehydrogenases, aldehyde dehydrogenases, carbonic anhydrases, aldo-keto reductases, esterases, epoxide hydrolases, and glutathione S-transferases were in abundance. This information bears great significance in gaining deeper insights on metabolic and bioactivation potential of both systemically as well as topically administered ophthalmic drugs as well as in the endobiotics metabolism within the eye. This investigation provides foundation for the subsequent targeted proteomic, genomic, and functional studies to explore the effects of sex, age, genotype and disease conditions across the populations.

A New Guidance for the Prediction of Hepatic Clearance in the Early Drug Discovery and Development from the in Vitro-to-in Vivo Extrapolation Method and an Approach for Exploring Whether an Albumin-Mediated Hepatic Uptake Phenomenon Could be Present Under in Vivo Conditions.

The in vitro-to-in vivo extrapolation (IVIVE) methods for predicting the hepatic clearance (CL) of drugs based on microsomal or hepatocyte data have certainly advanced; however, there is still place for improving the extrapolations from in vitro assays containing no plasma proteins. Accordingly, there is a discussion on whether the free drug hypothesis or an albumin (ALB)-mediated hepatic uptake phenomenon is the best scaling method. Therefore, the objectives of this study were to guide the prediction of CL and to diagnose which scaling method between the free drug hypothesis and ALB-mediated uptake could be more accurate; this, irrespective of the mechanism(s) governing CL if the drugs can get to the hepatocyte membrane. The analysis of several datasets demonstrated that almost all values of CL in vivo fall within the two calculated values of CL use as boundaries from: 1) the free drug hypothesis, and 2) ALB-mediated uptake. The average value from these two CL boundaries predicted the CL in vivo with an incredible accuracy. Validating these boundaries in preclinical species prior going to human as well as considering the fractional binding in plasma increased the accuracy. Overall, this study is another step towards guiding the CL prediction in drug discovery and development.

Application of HepG2/C3A liver spheroids as a model system for genotoxicity studies

HepG2 cells continue to be a valuable tool in early drug discovery and pharmaceutical development. In the current study we develop a 3D in vitro liver model, using HepG2/C3A cells that is predictive of human genotoxic exposure. HepG2/C3A cells cultured for 7-days in agarose-coated microplates formed spheroids which were uniform in shape and had well defined outer perimeters and no evidence of a hypoxic core. Quantitative real-time-PCR analysis showed statistically significant transcriptional upregulation of xenobiotic metabolising genes (CYP1A1, CYP1A2, UG1A1, UGT1A3, UGT1A6, EPHX, NAT2) and genes linked to liver function (ALB, CAR) in 3D cultures. In response to three model pro-genotoxicants: benzo[a]pyrene, amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and 2-aminoanthracene (2-AA), we observed further transcriptional upregulation of xenobiotic metabolising genes (CYP1A1, CYP1A2, NAT1/2, SULT1A2, UGT1A1, UGT1A3) compared to untreated spheroids. Consistent with this, spheroids were more sensitive than 2D monolayers to compound induced single- and double- stranded DNA-damage as assessed by the comet assay and γH2AX phosphorylation respectively. In contrast, levels of DNA-damage induced by the direct acting mutagen 4-nitroquinoline N-oxide (4NQO) was the same in spheroids and monolayers. In support of the enhanced genotoxic response in spheroids we also observed transcriptional upregulation of genes relating to DNA-damage and cellular stress response (e.g. GADD45A and CDKN1A) in spheroids. In conclusion, HepG2/C3A 3D spheroids are a sensitive model for in vitro genotoxicity assessment with potential applications in early stage drug development.

A Novel High-Throughput FLIPR Tetra-Based Method for Capturing Highly Confluent Kinetic Data for Structure-Kinetic Relationship Guided Early Drug Discovery.

Target engagement by small molecules is necessary for producing a physiological outcome. In the past, a lot of emphasis was placed on understanding the thermodynamics of such interactions to guide structure-activity relationships. It is becoming clearer, however, that understanding the kinetics of the interaction between a small-molecule inhibitor and the biological target [structure-kinetic relationship (SKR)] is critical for selection of the optimum candidate drug molecule for clinical trial. However, the acquisition of kinetic data in a high-throughput manner using traditional methods can be labor intensive, limiting the number of molecules that can be tested. As a result, in-depth kinetic studies are often carried out on only a small number of compounds, and usually at a later stage in the drug discovery process. Fundamentally, kinetic data should be used to drive key decisions much earlier in the drug discovery process, but the throughput limitations of traditional methods preclude this. A major limitation that hampers acquisition of high-throughput kinetic data is the technical challenge in collecting substantially confluent data points for accurate parameter estimation from time course analysis. Here, we describe the use of the fluorescent imaging plate reader (FLIPR), a charge-coupled device (CCD) camera technology, as a potential high-throughput tool for generating biochemical kinetic data with smaller time intervals. Subsequent to the design and optimization of the assay, we demonstrate the collection of highly confluent time-course data for various kinase protein targets with reasonable throughput to enable SKR-guided medicinal chemistry. We select kinase target 1 as a special case study with covalent inhibition, and demonstrate methods for rapid and detailed analysis of the resultant kinetic data for parameter estimation. In conclusion, this approach has the potential to enable rapid kinetic studies to be carried out on hundreds of compounds per week and drive project decisions with kinetic data at an early stage in drug discovery.

Establishment of inhibitor screening and validation system for tryptophanyl tRNA synthetase using surface plasmon resonance

With the increase in throughput and sensitivity, biophysical technology has become a major component of the early drug discovery phase. Surface plasmon resonance technology (SPR) is one of the most widely used biophysical technologies. It has the advantages of circumventing labeling, molecular weight limitations, and neglect of low affinity interactions, etc., and provides a robust platform for hit to lead discovery and optimization. Here, we successfully established a reliable and repeatable tryptophanyl tRNA synthetase (TrpRS) SPR high-throughput screening and validation system by optimizing the TrpRS tag, TrpRS immobilization methodology, and the buffer conditions. When TrpRS was immobilized on Streptavidin (SA) sensor chip, the substrate competitive inhibitor indolmycin exhibited the best binding affinity in HBS-P (10 mM HEPES, 150 mM NaCl, 0.05% surfactant P-20, pH 7.4), 1 mM ATP and MgCl2, with a KD (dissociation equilibrium constant) value of 0.6 ± 0.1 μM. The Z-factor values determined in the screening assays were all larger than 0.9. We hope that our proposed research ideas and methods may provide a scientific basis for establishing SPR analysis of other drug targets, accelerate the discovery and optimization of target lead compounds, and assist the clinical application of next-generation drugs.

Establishment of a Robust Platform for Induced Pluripotent Stem Cell Research Using Maholo LabDroid.

Induced pluripotent stem cells (iPSCs) are attractive for use in early drug discovery because they can differentiate into any cell type. Maintenance cultures and differentiation processes for iPSCs, however, require a high level of technical expertise. To overcome this problem, technological developments such as enhanced automation are necessary to replace manual operation. In addition, a robot system with the flexibility and expandability to carry out maintenance culture and each of the required differentiation processes would also be important. In this study, we established a platform to enable the multiple processes required for iPSC experiments using the Maholo LabDroid, which is a humanoid robotic system with excellent reproducibility and flexibility. The accuracy and robustness of Maholo LabDroid enabled us to cultivate undifferentiated iPSCs for 63 days while maintaining their ability to differentiate into the three embryonic germ layers. Maholo LabDroid maintained and harvested iPSCs in six-well plates, then seeded them into 96-well plates, induced differentiation, and implemented immunocytochemistry. As a result, Maholo LabDroid was confirmed to be able to perform the processes required for myogenic differentiation of iPSCs isolated from a patient with muscular disease and achieved a high differentiation rate with a coefficient of variation (CV) <10% in the first trial. Furthermore, the expandability and flexibility of Maholo LabDroid allowed us to experiment with multiple cell lines simultaneously.

How to Develop and Prove High-Efficiency Selection of Ligands from Oligonucleotide Libraries: A Universal Framework for Aptamers and DNA-Encoded Small-Molecule Ligands.

Screening molecular libraries for ligands capable of binding proteins is widely used for hit identification in the early drug discovery process. Oligonucleotide libraries provide a very high diversity of compounds, while the combination of the polymerase chain reaction and DNA sequencing allow the identification of ligands in low copy numbers selected from such libraries. Ligand selection from oligonucleotide libraries requires mixing the library with the target followed by the physical separation of the ligand-target complexes from the unbound library. Cumulatively, the low abundance of ligands in the library and the low efficiency of available separation methods necessitate multiple consecutive rounds of partitioning. Multiple rounds of inefficient partitioning make the selection process ineffective and prone to failures. There are continuing efforts to develop a separation method capable of reliably generating a pure pool of ligands in a single round of partitioning; however, none of the proposed methods for single-round selection have been universally adopted. Our analysis revealed that the developers' efforts are disconnected from each other and hindered by the lack of quantitative criteria of selection quality assessment. Here, we present a formalism that describes single-round selection mathematically and provides parameters for quantitative characterization of selection quality. We use this formalism to define a universal strategy for development and validation of single-round selection methods. Finally, we analyze the existing partitioning methods, the published single-round selection reports, and some pertinent practical considerations through the prism of this formalism. This formalism is not an experimental protocol but a framework for correct development of experimental protocols. While single-round selection is not a goal by itself and may not always suffice selection of good-quality ligands, our work will help developers of highly efficient selection approaches to consolidate their efforts under an umbrella of universal quantitative criteria of method development and assessment.

Universal Automated Immunoaffinity Purification-CE-MS Platform for Accelerating Next Generation Biologic Design.

As the pharmaceutical industry places greater emphasis on pairing biological pathways with appropriate therapeutic intervention, an increase in the use of biologic drugs has emerged. With increasing complexity of biotherapeutics, absorption, distribution, metabolism, and excretion (ADME) studies have also become increasingly complex. The characterization of ADME properties is critical to tuning the pharmacokinetic profiles of next generation biologics (NGBs). The knowledge of the fate of a drug is essential for the enhancement of the design processes, elongation of exposure at the desired site of action, and achieving efficacy with minimum toxicity. In vivo proteolytic cleavage of biotherapeutics may lead to undesirable in vivo properties, such as rapid clearance, low bioavailability, and loss of pharmacodynamic effect. All of these may affect drug efficacy and/or generate safety concerns through increases in immunogenicity or off-target toxicity. The work herein describes the development of a robust, fully automated immunoaffinity purification (IA)-capillary electrophoresis-mass spectrometry (CE-MS) workflow. The reagents were carefully optimized to maximize isolation yields while minimizing the number of experimental steps to analytical results. The result is the development of a comprehensive integrated platform for the characterization of a wide range of biotherapeutics, including peptibodies, monoclonal antibodies, and bispecific antibodies. Empowered by this automated IA-CE-MS approach, implementing biotransformation studies at an early drug discovery stage can speed up the drug development process.

Model-based Target Pharmacology Assessment (mTPA): An Approach Using PBPK/PD Modeling and Machine Learning to Design Medicinal Chemistry and DMPK Strategies in Early Drug Discovery.

The optimal pharmacokinetic (PK) required for a drug candidate to elicit efficacy is highly dependent on the targeted pharmacology, a relationship that is often not well characterized during early phases of drug discovery. Generic assumptions around PK and potency risk misguiding screening and compound design toward nonoptimal absorption, distribution, metabolism, and excretion (ADME) or molecular properties and ultimately may increase attrition as well as hit-to-lead and lead optimization timelines. The present work introduces model-based target pharmacology assessment (mTPA), a computational approach combining physiologically based pharmacokinetic/pharmacodynamic (PBPK/PD) modeling, sensitivity analysis, and machine learning (ML) to elucidate the optimal combination of PK, potency, and ADME specific for the targeted pharmacology. Examples using frequently encountered PK/PD relationships are presented to illustrate its application, and the utility and benefits of deploying such an approach to guide early discovery efforts are discussed.

Rifampicin as an example of beyond-rule-of-5 compound: Ionization beyond water and lipophilicity beyond octanol/water

Ionization and lipophilicity in early drug discovery are commonly characterized in water and octanol/water, respectively and thus do not consider the non-polar features of the biomembrane core. This is particularly limiting for bRo5 compounds which may adapt their properties (e.g. ionization and lipophilicity) to the environment. In this paper we used experimental methods to characterize rifampicin for its ionization properties in various water/cosolvent mixtures and in pure MeCN and its lipophilicity in octanol/water and toluene/water systems. Moreover, we also measured log k’80 PLRP-S, a chromatographic index of lipophilicity in non-polar media. Results show that the existence domain of neutral rifampicin is limited compared to the zwitterion, but the lipophilic cationic species is extremely relevant in non-polar environments.

Small-Molecule Degraders beyond PROTACs-Challenges and Opportunities.

Targeted protein degradation (TPD) is a recent strategy, utilizing the cell's proteostasis machinery to deplete specific proteins. This represents a paradigm shift in early drug discovery, away from occupancy-driven to event-driven mechanisms.Recent efforts have focused on the development of proteolysis-targeting chimeras (PROTACs). These heterobifunctional molecules combine a target-specific binding moiety linked to an E3 ligase ligand and trigger selective ubiquitination of the target protein, marking it for proteasomal degradation. While these molecules can be highly efficacious, they generally have unfavorable physicochemical properties due to their large size.In contrast, smaller molecules that induce degradation could represent an attractive, simple option to overcoming the limitations of both traditional modulators and PROTACs. These molecules may have a range of mechanisms: recruitment of an E3 ligase (molecular glues), introduction of hydrophobic areas, or inducing local unfolding, each of which triggers degradation.We recently completed a high-throughput screen of 111,000 compounds in a cellular HiBiT assay in an effort to identify such molecules. Preliminary analysis indicates that we have been able to identify alternative small-molecule degraders. We highlight methods for triage, characterization, selectivity, and mode of action. In summary, we believe that these types of small-molecule degraders, which may possibly have more acceptable physicochemical properties than the inherently larger heterobifunctional molecules, are an exciting approach for inducing TPD, and we illustrate that a general screening approach can be successful in identifying useful start points for developing such molecules.

Fast Identification of Adverse Drug Reactions (ADRs) of Digestive and Nervous Systems of Organic Drugs by In Silico Models.

This study aimed to discover concurrences of adverse drug reactions (ADRs) and derive models of the most frequent items of ADRs based on the SIDER database, which included 1430 marketed drugs and 5868 ADRs. First, common ADRs of organic drugs were manually reclassified according to side effects in the human system and followed by an association rule analysis, which found ADRs of digestive and nervous systems often occurred at the same time with a good association rule. Then, three algorithms, linear discriminant analysis (LDA), support vector machine (SVM) and deep learning, were used to derive models of ADRs of digestive and nervous systems based on 497 organic monomer drugs and to identify key structural features in defining these ADRs. The statistical results indicated that these kinds of QSAR models were good tools for screening ADRs of digestive and nervous systems, which gave the ROC AUC values of 81.5%, 98.9%, 91.5%, 69.5%, 78.4% and 78.8%, respectively. Then, these models were applied to investigate ADRs of 1536 organic compounds with four phase and zero rule-of-five (RO5) violations from the ChEMBL database. Based on the consensus ADRs’ predictions of models, 58.1% and 42.6% of compounds were predicted to cause these two ADRs, respectively, indicating the significance of initial assessment of ADRs in early drug discovery.