The (Re)-Evolution of Quantitative Structure-Activity Relationship (QSAR) Studies Propelled by the Surge of Machine Learning Methods

Strategic lead compound design and development utilizing computer-aided drug discovery (CADD) to address herbicide-resistant Phalaris minor in wheat fields

Wheat (Triticum aestivum) is a vital cereal crop and a staple food source worldwide. However, wheat grain productivity has significantly declined as a consequence of infestations by Phalaris minor. Traditional weed control methods have proven inadequate owing to the physiological similarities between P. minor and wheat during early growth stages. Consequently, farmers have turned to herbicides, targeting acetyl-CoA carboxylase (ACCase), acetolactate synthase (ALS) and photosystem II (PSII). Isoproturon targeting PSII was introduced in mid-1970s, to manage P. minor infestations. Despite their effectiveness, the repetitive use of these herbicides has led to the development of herbicide-resistant P. minor biotypes, posing a significant challenge to wheat productivity. To address this issue, there is a pressing need for innovative weed management strategies and the discovery of novel herbicide molecules. The integration of computer-aided drug discovery (CADD) techniques has emerged as a promising approach in herbicide research, that facilitates the identification of herbicide targets and enables the screening of large chemical libraries for potential herbicide-like molecules. By employing techniques such as homology modelling, molecular docking, molecular dynamics simulation and pharmacophore modelling, CADD has become a rapid and cost-effective medium to accelerate the herbicide discovery process significantly. This approach not only reduces the dependency on traditional experimental methods, but also enhances the precision and efficacy of herbicide development. This article underscores the critical role of bioinformatics and CADD in developing next-generation herbicides, offering new hope for sustainable weed management and improved wheat cultivation practices. © 2024 Society of Chemical Industry.

Artificial Intelligence for the Discovery of Safe and Effective Flame Retardants

Organophosphorus flame retardants (OPFRs) are important chemical additives that are used in commercial products. However, owing to increasing health concerns, the discovery of new OPFRs has become imperative. Herein, we propose an explainable artificial intelligence-assisted product design (AIPD) methodological framework for screening novel, safe, and effective OPFRs. Using a deep neural network, we established a flame retardancy prediction model with an accuracy of 0.90. Employing the SHapley Additive exPlanations approach, we have identified the Morgan 507 (P═N connected to a benzene ring) and 114 (quaternary carbon) substructures as promoting units in flame retardancy. Subsequently, approximately 600 compounds were selected as OPFR candidates from the ZINC database. Further refinement was achieved through a comprehensive scoring system that incorporated absorption, toxicity, and persistence, thereby yielding six prospective candidates. We experimentally validated these candidates and identified compound Z2 as a promising candidate, which was not toxic to zebrafish embryos. Our methodological framework leverages AIPD to effectively guide the discovery of novel flame retardants, significantly reducing both developmental time and costs.

Multitarget Natural Compounds for Ischemic Stroke Treatment: Integration of Deep Learning Prediction and Experimental Validation

Ischemic stroke's complex pathophysiology demands therapeutic approaches targeting multiple pathways simultaneously, yet current treatments remain limited. We developed an innovative drug discovery pipeline combining a deep learning approach with experimental validation to identify natural compounds with comprehensive neuroprotective properties. Our computational framework integrated SELFormer, a transformer-based deep learning model, and multiple deep learning algorithms to predict NC bioactivity against seven crucial stroke-related targets (ACE, GLA, MMP9, NPFFR2, PDE4D, and eNOS). The pipeline encompassed IC50 predictions, clustering analysis, quantitative structure-activity relationship (QSAR) modeling, and uniform manifold approximation and projection (UMAP)-based bioactivity profiling followed by molecular docking studies and experimental validation. Analysis revealed six distinct NC clusters with unique molecular signatures. UMAP projection identified 11 medium-activity (6 < pIC50 ≤ 7) and 57 high-activity (pIC50 > 7) compounds, with molecular docking confirming strong correlations between binding energies and predicted pIC50 values. In vitro studies using NGF-differentiated PC12 cells under oxygen-glucose deprivation demonstrated significant neuroprotective effects of four high-activity compounds: feruloyl glucose, l-hydroxy-l-tryptophan, mulberrin, and ellagic acid. These compounds enhanced cell viability, reduced acetylcholinesterase activity and lipid peroxidation, suppressed TNF-α expression, and upregulated BDNF mRNA levels. Notably, mulberrin and ellagic acid showed superior efficacy in modulating oxidative stress, inflammation, and neurotrophic signaling. This study establishes a robust deep learning-driven framework for identifying multitarget natural therapeutics for ischemic stroke. The validated compounds, particularly mulberrin and ellagic acid, are promising for stroke treatment development. Our findings demonstrate the effectiveness of integrating computational prediction with experimental validation in accelerating drug discovery for complex neurological disorders.

Predicting the Mutagenic Activity of Nitroaromatics Using Conceptual Density Functional Theory Descriptors and Explainable No-Code Machine Learning Approaches

Nitroaromatic compounds (NAs) are widely used in industrial applications but pose significant genotoxic risks, necessitating accurate mutagenicity prediction for chemical safety assessments. This study integrates conceptual density functional theory (CDFT) descriptors with explainable no-code machine learning (ML) models to predict NA mutagenicity based on Ames test results. Following OECD QSAR guidelines, feature selection and model development were performed using decision-tree-based algorithms (Random Tree, JCHAID*, SPAARC) and multilayer perceptrons (MLPs). These models exhibited high predictive accuracy (internal: >80%, κ = 0.21-0.37; external: ∼90%, κ = 0.41-0.62) with strong interpretability. The study also explores the role of metabolic activation and aqueous-phase descriptors, evaluating a novel electronic analog to LogP (LogQP) to assess hydrophobicity-mutagenicity relationships. Results demonstrate that aqueous-phase electronic properties and electrophilicity descriptors outperform vacuum-based methods in mutagenicity prediction. The combination of CDFT descriptors with shallow ML models proves to be a robust, interpretable, and accessible framework for predictive toxicology. This approach enhances chemical risk assessment and bridges computational chemistry with toxicology for regulatory applications.

Biological activity of natural 2-quinolinones

While natural products have undeniably played a crucial role in drug discovery, challenges such as limited availability and complex synthesis methods have hindered the identification of lead compounds. At the core of numerous natural and synthetic compounds, each displaying distinct biological behaviours, lies the foundational structure of 2-quinolinone. Compounds with this structural motif exhibit a broad range of effects in different tissues. Furthermore, specific members showcase therapeutic potential for various disorders. Despite the significance of these compounds, the current review literature has not provided a comprehensive overview, underscoring the essential contribution of this article in exploring their biological functions. This study examines the biological activity of selected 2-quinolinone alkaloids across diverse organisms, unveiling their potential as a source of innovative bioactive natural products.

Design Principles From Natural Olfaction for Electronic Noses

Natural olfactory systems possess remarkable sensitivity and precision beyond what is currently achievable by engineered gas sensors. Unlike their artificial counterparts, noses are capable of distinguishing scents associated with mixtures of volatile molecules in complex, typically fluctuating environments and can adapt to changes. This perspective examines the multifaceted biological principles that provide olfactory systems their discriminatory prowess, and how these ideas can be ported to the design of electronic noses for substantial improvements in performance across metrics such as sensitivity and ability to speciate chemical mixtures. The topics examined herein include the fluid dynamics of odorants in natural channels; specificity and kinetics of odorant interactions with olfactory receptors and mucus linings; complex signal processing that spatiotemporally encodes physicochemical properties of odorants; active sampling techniques, like biological sniffing and nose repositioning; biological priming; and molecular chaperoning. Each of these components of natural olfactory systems are systmatically investigated, as to how they have been or can be applied to electronic noses. While not all artificial sensors can employ these strategies simultaneously, integrating a subset of bioinspired principles can address issues like sensitivity, drift, and poor selectivity, offering advancements in many sectors such as environmental monitoring, industrial safety, and disease diagnostics.

QSAR, Antimicrobial, and Antiproliferative Study of (R/S)-2-Thioxo-3,4-dihydropyrimidine-5-carboxanilides

Owing to the significant contribution of three-dimensional (3D) field-based QSAR toward hit optimization and accurately predicting the activities of small molecules, herein, the 3D-QSAR, in vitro antimicrobial, molecular docking, and pharmacophore modeling studies of all the isolated (R/S)-2-thioxo-DHPM-5-carboxanilides exhibiting antimicrobial activity were carried out. The screening process was performed using 46 compounds, and the best-scoring model with the top statistical values was considered for bacterial and fungal targets Bacillus subtilis and Candida albicans. As a result of 3D-QSAR analysis, compound 4v-(S)- and 4v-(R)-isomers were found to be more potent compared to the standard drugs tetracycline and fluconazole, respectively. Furthermore, the enantiomerically pure isomers 4q, 4d', 4n, 4f', 4v, 4q', 4c, and 4p' were found to be more potent than tetracycline and fluconazole to inhibit the bacterial and fungal growth against B. subtilis, Salinivibrio proteolyticus, C. albicans, and Aspergillus niger, respectively. Molecular docking analysis shows that with the glide score of -10.261 kcal/mol, 4v-(R)-isomer was found to be more potent against the fungal target C. albicans and may target the 14-α demethylase than fluconazole. Furthermore, all compounds' antiproliferative activity results showed that 4o' exhibited GI50 values between 8.8 and 34 μM against six solid tumor cell lines. Following the greater potential of 4o' toward the HeLa cell line, its kinetics study and live cell imaging were carried out. These outcomes highlight the acceptance and safety as well as the potential of compounds as effective antiproliferative and antifungal agents.

Nanomaterials and methods for cancer therapy: 2D materials, biomolecules, and molecular dynamics simulations

This review explores the potential of biomolecule-based nanomaterials, i.e., protein, peptide, nucleic acid, and polysaccharide-based nanomaterials, in cancer nanomedicine. It highlights the wide range of design possibilities for creating multifunctional nanomedicines using these biomolecule-based nanomaterials. This review also analyzes the primary obstacles in cancer nanomedicine that can be resolved through the usage of nanomaterials based on biomolecules. It also examines the unique in vivo characteristics, programmability, and biological functionalities of these biomolecule-based nanomaterials. This summary outlines the most recent advancements in the development of two-dimensional semiconductor-based nanomaterials for cancer theranostic purposes. It focuses on the latest developments in molecular simulations and modelling to provide a clear understanding of important uses, techniques, and concepts of nanomaterials in drug delivery and synthesis processes. Finally, the review addresses the challenges in molecular simulations, and generating, analyzing, and developing biomolecule-based and two-dimensional semiconductor-based nanomaterials, and highlights the barriers that must be overcome to facilitate their application in clinical settings.

Synergizing Machine Learning, Conceptual Density Functional Theory, and Biochemistry: No-Code Explainable Predictive Models for Mutagenicity in Aromatic Amines

This study synergizes machine learning (ML) with conceptual density functional theory (CDFT) to develop OECD-compliant predictive models for the mutagenic activity of aromatic amines (AAs) with a fully No-Code methodology using a comprehensive data set of 251 AAs, Leave-One-Out-Cross-Validation (LOOCV), and three distinct data splits. Our research employs the GFN2-xTB method, known for its robustness and speed, to compute descriptors for procarcinogens and their activated metabolites in vacuum and aqueous phases. We evaluate the effectiveness of different theoretical definitions of electrophilicity within CDFT, namely, PSL, GCV, and CDP schemes, and the newly introduced Log QP descriptor to approximate Log P information. SPAARC, RandomTree, and JCHAID* ML methods were used to build explainable predictive models with highly robust internal validation (Avg. Correct Classifications = 76% and Avg. Kappa = 0.29) and external validation (Avg. Correct Classifications = 79% and Avg. Kappa = 0.33) metrics, and the results were compared to those of a two hidden layer Multilayer Perceptron. The results indicate that the second CDP definition for the electrophilicity in both vacuum and aqueous phases and also the newly presented Log QP descriptors are the most important ones for predicting the mutagenic activity of AA (namely ω+VacCDP2+, ω+AqCDP2+, and LogQP1+Vac, respectively). The results indicate that metabolic activation, aqueous solvent properties, and the CDP electrophilicity schemes and Log QP should be considered when building predictive models for the mutagenic activity of AA. This study offers a replicable, No-Code approach to QSAR research, making high-level ML and CDFT applications accessible to a broader audience. Future work will expand these methods to other compound families, enhancing predictive capabilities in the study of mutagenic activities and other biological phenomena.

Artificial Intelligence-Driven Computational Approaches in the Development of Anticancer Drugs

The integration of AI has revolutionized cancer drug development, transforming the landscape of drug discovery through sophisticated computational techniques. AI-powered models and algorithms have enhanced computer-aided drug design (CADD), offering unprecedented precision in identifying potential anticancer compounds. Traditionally, cancer drug design has been a complex, resource-intensive process, but AI introduces new opportunities to accelerate discovery, reduce costs, and optimize efficiency. This manuscript delves into the transformative applications of AI-driven methodologies in predicting and developing anticancer drugs, critically evaluating their potential to reshape the future of cancer therapeutics while addressing their challenges and limitations.

Advancing toxicity studies of per- and poly-fluoroalkyl substances (pfass) through machine learning: Models, mechanisms, and future directions

Perfluorinated and perfluoroalkyl substances (PFASs), encompassing a vast array of isomeric chemicals, are recognized as typical emerging contaminants with direct or potential impacts on human health and the ecological environment. With the complex and elusive toxicological profiles of PFASs, machine learning (ML) has been increasingly employed in their toxicity studies due to its proficiency in prediction and data analytics. This integration is poised to become a predominant trend in environmental toxicology, propelled by the swift advancements in computational technology. This review diligently examines the literature to encapsulate the varied objectives of employing ML in the toxicity studies of PFASs: (1) Utilizing ML to establish Quantitative Structure-Activity Relationship (QSAR) models for PFASs with diverse toxicity endpoints, facilitating the targeted toxicity prediction of unidentified PFASs; (2) Investigating and substantiating the Adverse Outcome Pathway (AOP) through the synergy of ML and traditional toxicological methods, with this refining the toxicity assessment framework for PFASs; (3) Dissecting and elucidating the features of established ML models to advance Open Research into the toxicity of PFASs, with a primary focus on determinants and mechanisms. The discourse extends to an in-depth examination of ML studies, segregating findings based on their distinct application trajectories. Given that ML represents a nascent paradigm within PFASs research, this review delineates the collective challenges encountered in the ML-mediated study of PFAS toxicity and proffers strategic guidance for ensuing investigations.

Employing Machine Learning-Based QSAR for Targeting Zika Virus NS3 Protease: Molecular Insights and Inhibitor Discovery

Zika virus infection is a mosquito-borne viral disease that has become a global health concern recently. Zika virus belongs to the Flavivirus genus and is primarily transmitted by Aedes mosquitoes. Prevention of Zika virus infection involves avoiding mosquito bites by using repellent, wearing protective clothing, and staying in screened areas, especially for pregnant women. Treatment focuses on managing symptoms with rest, fluids, and acetaminophen, with close monitoring for pregnant women. Currently, there is no specific antiviral treatment or vaccine for the Zika virus, highlighting the importance of prevention strategies to control its spread. Therefore, in this study, the Zika virus non-structural protein NS3 was targeted to inhibit Zika infection by identifying the novel inhibitor through an in silico approach. Here, 2864 natural compounds were screened using a machine learning-based QSAR model, and later docking was performed to select the potential target. Subsequently, Tanimoto similarity and clustering were performed to obtain the potential target. The three most potential compounds were obtained: (a) 5297, (b) 432449, and (c) 85137543. The protein-ligand complex's stability and flexibility were then investigated by dynamic modelling. The 300 ns simulation showed that 5297 exhibited the steadiest deviation and constant creation of hydrogen bonds. Compared to the other compounds, 5297 demonstrated a superior binding free energy (ΔG = -20.81 kcal/mol) with the protein when the MM/GBSA technique was used. The study determined that 5297 showed significant therapeutic potential and justifies further experimental investigation as a possible inhibitor of the NS2B-NS3 protease target implicated in Zika virus infection.

HydraScreen: A Generalizable Structure-Based Deep Learning Approach to Drug Discovery

We propose HydraScreen, a deep-learning framework for safe and robust accelerated drug discovery. HydraScreen utilizes a state-of-the-art 3D convolutional neural network designed for the effective representation of molecular structures and interactions in protein-ligand binding. We designed an end-to-end pipeline for high-throughput screening and lead optimization, targeting applications in structure-based drug design. We assessed our approach using established public benchmarks based on the CASF-2016 core set, achieving top-tier results in affinity and pose prediction (Pearson's r = 0.86, RMSE = 1.15, Top-1 = 0.95). We introduced a novel approach for interaction profiling, aimed at detecting potential biases within both the model and data sets. This approach not only enhanced interpretability but also reinforced the impartiality of our methodology. Finally, we demonstrated HydraScreen's ability to generalize effectively across novel proteins and ligands through a temporal split. We also provide insights into potential avenues for future development aimed at enhancing the robustness of machine learning scoring functions. HydraScreen (accessible at http://hydrascreen.ro5.ai/paper) provides a user-friendly GUI and a public API, facilitating the easy-access assessment of protein-ligand complexes.

Advances in artificial intelligence for drug delivery and development: A comprehensive review

Artificial intelligence (AI) has emerged as a powerful tool to revolutionize the healthcare sector, including drug delivery and development. This review explores the current and future applications of AI in the pharmaceutical industry, focusing on drug delivery and development. It covers various aspects such as smart drug delivery networks, sensors, drug repurposing, statistical modeling, and simulation of biotechnological and biological systems. The integration of AI with nanotechnologies and nanomedicines is also examined. AI offers significant advancements in drug discovery by efficiently identifying compounds, validating drug targets, streamlining drug structures, and prioritizing response templates. Techniques like data mining, multitask learning, and high-throughput screening contribute to better drug discovery and development innovations. The review discusses AI applications in drug formulation and delivery, clinical trials, drug safety, and pharmacovigilance. It addresses regulatory considerations and challenges associated with AI in pharmaceuticals, including privacy, data security, and interpretability of AI models. The review concludes with future perspectives, highlighting emerging trends, addressing limitations and biases in AI models, and emphasizing the importance of collaboration and knowledge sharing. It provides a comprehensive overview of AI's potential to transform the pharmaceutical industry and improve patient care while identifying further research and development areas.

Decoding Nanomaterial-Biosystem Interactions through Machine Learning

The interactions between biosystems and nanomaterials regulate most of their theranostic and nanomedicine applications. These nanomaterial-biosystem interactions are highly complex and influenced by a number of entangled factors, including but not limited to the physicochemical features of nanomaterials, the types and characteristics of the interacting biosystems, and the properties of the surrounding microenvironments. Over the years, different experimental approaches coupled with computational modeling have revealed important insights into these interactions, although many outstanding questions remain unanswered. The emergence of machine learning has provided a timely and unique opportunity to revisit nanomaterial-biosystem interactions and to further push the boundary of this field. This minireview highlights the development and use of machine learning to decode nanomaterial-biosystem interactions and provides our perspectives on the current challenges and potential opportunities in this field.

Unlocking the Potential of High-Quality Dopamine Transporter Pharmacological Data: Advancing Robust Machine Learning-Based QSAR Modeling

The dopamine transporter (DAT) plays a critical role in the central nervous system and has been implicated in numerous psychiatric disorders. The ligand-based approaches are instrumental to decipher the structure-activity relationship (SAR) of DAT ligands, especially the quantitative SAR (QSAR) modeling. By gathering and analyzing data from literature and databases, we systematically assemble a diverse range of ligands binding to DAT, aiming to discern the general features of DAT ligands and uncover the chemical space for potential novel DAT ligand scaffolds. The aggregation of DAT pharmacological activity data, particularly from databases like ChEMBL, provides a foundation for constructing robust QSAR models. The compilation and meticulous filtering of these data, establishing high-quality training datasets with specific divisions of pharmacological assays and data types, along with the application of QSAR modeling, prove to be a promising strategy for navigating the pertinent chemical space. Through a systematic comparison of DAT QSAR models using training datasets from various ChEMBL releases, we underscore the positive impact of enhanced data set quality and increased data set size on the predictive power of DAT QSAR models.

A breast cancer-specific combinational QSAR model development using machine learning and deep learning approaches

Breast cancer is the most prevalent and heterogeneous form of cancer affecting women worldwide. Various therapeutic strategies are in practice based on the extent of disease spread, such as surgery, chemotherapy, radiotherapy, and immunotherapy. Combinational therapy is another strategy that has proven to be effective in controlling cancer progression. Administration of Anchor drug, a well-established primary therapeutic agent with known efficacy for specific targets, with Library drug, a supplementary drug to enhance the efficacy of anchor drugs and broaden the therapeutic approach. Our work focused on harnessing regression-based Machine learning (ML) and deep learning (DL) algorithms to develop a structure-activity relationship between the molecular descriptors of drug pairs and their combined biological activity through a QSAR (Quantitative structure-activity relationship) model. 11 popularly known machine learning and deep learning algorithms were used to develop QSAR models. A total of 52 breast cancer cell lines, 25 anchor drugs, and 51 library drugs were considered in developing the QSAR model. It was observed that Deep Neural Networks (DNNs) achieved an impressive R2 (Coefficient of Determination) of 0.94, with an RMSE (Root Mean Square Error) value of 0.255, making it the most effective algorithm for developing a structure-activity relationship with strong generalization capabilities. In conclusion, applying combinational therapy alongside ML and DL techniques represents a promising approach to combating breast cancer.

3D QSAR study on substituted 1, 2, 4 triazole derivatives as anticancer agents by kNN MFA approach

Background and objectives

Researchers have recently focused on the biological and synthetic effects of 1, 2, and 4-triazole fused heterocyclic molecules because they have tremendous medicinal value. The objective of the present study was to carry out the 3D QSAR evaluation on the substituted 1,2, and 4 triazole derivatives for anticancer potential using k-Nearest Neighbor-Molecular Field Analysis (kNN-MFA) method.

Methods

Using the molecular design suite, a three-dimensional quantitative structure-activity relationship (3D-QSAR) analysis was undertaken on a series of 4-amino-5-(pyridin3yl)-4H-1, 2, and 4-triazole-3-thiol anticancer drugs (Vlife MDS). This study used a genetic algorithm and a manual selection approach on 20 substituted 1, 2, and 4-triazole derivatives. Based on the genetic algorithm (GA), the 3D-QSAR model was generated. Statistical significance and predictive capacity were evaluated using internal and external validation.

Results

The most significant model has a correlation coefficient of 0.9334 (squared correlation coefficient r2 = 0.8713), showing that biological activity and descriptors have a strong relationship. The model exhibited internal predictivity of 74.45 percent (q2 = 0.2129), external predictivity of 81.09 percent (pred r2 = 0.8417), and the smallest error term for the predictive correlation coefficient (pred r2se = 0.1255). The model revealed steric (S 1047--0.0780--0.0451S 927) and electrostatic (E 1002) data points that contribute remarkably to anticancer activity. A molecular field study demonstrates a link between the structural features of substituted triazole derivatives and their activities.

Conclusion

The good-to-moderate anticancer potential of compounds confirms the significant pharmacological role of 1,2,4-triazole derivatives. These results could lead to the identification of potential chemical compounds with optimal anticancer activity and minimal side effects.

Prospection of Peptide Inhibitors of Thrombin from Diverse Origins Using a Machine Learning Pipeline

Thrombin is a key enzyme involved in the development and progression of many cardiovascular diseases. Direct thrombin inhibitors (DTIs), with their minimum off-target effects and immediacy of action, have greatly improved the treatment of these diseases. However, the risk of bleeding, pharmacokinetic issues, and thrombotic complications remain major concerns. In an effort to increase the effectiveness of the DTI discovery pipeline, we developed a two-stage machine learning pipeline to identify and rank peptide sequences based on their effective thrombin inhibitory potential. The positive dataset for our model consisted of thrombin inhibitor peptides and their binding affinities (KI) curated from published literature, and the negative dataset consisted of peptides with no known thrombin inhibitory or related activity. The first stage of the model identified thrombin inhibitory sequences with Matthew's Correlation Coefficient (MCC) of 83.6%. The second stage of the model, which covers an eight-order of magnitude range in KI values, predicted the binding affinity of new sequences with a log room mean square error (RMSE) of 1.114. These models also revealed physicochemical and structural characteristics that are hidden but unique to thrombin inhibitor peptides. Using the model, we classified more than 10 million peptides from diverse sources and identified unique short peptide sequences (<15 aa) of interest, based on their predicted KI. Based on the binding energies of the interaction of the peptide with thrombin, we identified a promising set of putative DTI candidates. The prediction pipeline is available on a web server.

Quantitative Systems Pharmacology and Machine Learning: A Match Made in Heaven or Hell?

As pharmaceutical development moves from early-stage in vitro experimentation to later in vivo and subsequent clinical trials, data and knowledge are acquired across multiple time and length scales, from the subcellular to whole patient cohort scale. Realizing the potential of this data for informing decision making in pharmaceutical development requires the individual and combined application of machine learning (ML) and mechanistic multiscale mathematical modeling approaches. Here we outline how these two approaches, both individually and in tandem, can be applied at different stages of the drug discovery and development pipeline to inform decision making compound development. The importance of discerning between knowledge and data are highlighted in informing the initial use of ML or mechanistic quantitative systems pharmacology (QSP) models. We discuss the application of sensitivity and structural identifiability analyses of QSP models in informing future experimental studies to which ML may be applied, as well as how ML approaches can be used to inform mechanistic model development. Relevant literature studies are highlighted and we close by discussing caveats regarding the application of each approach in an age of constant data acquisition. SIGNIFICANCE STATEMENT: We consider when best to apply machine learning (ML) and mechanistic quantitative systems pharmacology (QSP) approaches in the context of the drug discovery and development pipeline. We discuss the importance of prior knowledge and data available for the system of interest and how this informs the individual and combined application of ML and QSP approaches at each stage of the pipeline.

First fully-automated AI/ML virtual screening cascade implemented at a drug discovery centre in Africa

Streamlined data-driven drug discovery remains challenging, especially in resource-limited settings. We present ZairaChem, an artificial intelligence (AI)- and machine learning (ML)-based tool for quantitative structure-activity/property relationship (QSAR/QSPR) modelling. ZairaChem is fully automated, requires low computational resources and works across a broad spectrum of datasets. We describe an end-to-end implementation at the H3D Centre, the leading integrated drug discovery unit in Africa, at which no prior AI/ML capabilities were available. By leveraging in-house data collected over a decade, we have developed a virtual screening cascade for malaria and tuberculosis drug discovery comprising 15 models for key decision-making assays ranging from whole-cell phenotypic screening and cytotoxicity to aqueous solubility, permeability, microsomal metabolic stability, cytochrome inhibition, and cardiotoxicity. We show how computational profiling of compounds, prior to synthesis and testing, can inform progression of frontrunner compounds at H3D. This project is a first-of-its-kind deployment at scale of AI/ML tools in a research centre operating in a low-resource setting.

Computational and biophysical methods for the discovery and optimization of covalent drugs

Drugs that act by covalently attaching to their targets have been used to treat human diseases for over a hundred years. However, the deliberate design of covalent drugs was discouraged due to concerns of toxicity and off-target effects. Recent successes in covalent drug discovery have sparked fresh interest in this field. New screening and testing methods aimed at covalent inhibitors can play pivotal roles in facilitating the discovery process. This feature article focuses on computational and biophysical advances originating from our labs over the past decade and how these approaches have contributed to the design of prolyl oligopeptidase (POP) and SARS-CoV-2 3CLpro covalent inhibitors.

Equivariant Graph Neural Networks for Toxicity Prediction

Predictive modeling of toxicity is a crucial step in the drug discovery pipeline. It can help filter out molecules with a high probability of failing in the early stages of de novo drug design. Thus, several machine learning (ML) models have been developed to predict the toxicity of molecules by combining classical ML techniques or deep neural networks with well-known molecular representations such as fingerprints or 2D graphs. But the more natural, accurate representation of molecules is expected to be defined in physical 3D space like in ab initio methods. Recent studies successfully used equivariant graph neural networks (EGNNs) for representation learning based on 3D structures to predict quantum-mechanical properties of molecules. Inspired by this, we investigated the performance of EGNNs to construct reliable ML models for toxicity prediction. We used the equivariant transformer (ET) model in TorchMD-NET for this. Eleven toxicity data sets taken from MoleculeNet, TDCommons, and ToxBenchmark have been considered to evaluate the capability of ET for toxicity prediction. Our results show that ET adequately learns 3D representations of molecules that can successfully correlate with toxicity activity, achieving good accuracies on most data sets comparable to state-of-the-art models. We also test a physicochemical property, namely, the total energy of a molecule, to inform the toxicity prediction with a physical prior. However, our work suggests that these two properties can not be related. We also provide an attention weight analysis for helping to understand the toxicity prediction in 3D space and thus increase the explainability of the ML model. In summary, our findings offer promising insights considering 3D geometry information via EGNNs and provide a straightforward way to integrate molecular conformers into ML-based pipelines for predicting and investigating toxicity prediction in physical space. We expect that in the future, especially for larger, more diverse data sets, EGNNs will be an essential tool in this domain.

Unlocking the Potential: Predicting Redox Behavior of Organic Molecules, from Linear Fits to Neural Networks

Redox-active organic molecules, i.e., molecules that can relatively easily accept and/or donate electrons, are ubiquitous in biology, chemical synthesis, and electronic and spintronic devices, such as solar cells and rechargeable batteries, etc. Choosing the best candidates from an essentially infinite chemical space for experimental testing in a target application requires efficient screening approaches. In this Review, we discuss modern in silico techniques for predicting reduction and oxidation potentials of organic molecules that go beyond conventional first-principles computations and thermodynamic cycles. Approaches ranging from simple linear fits based on molecular orbital energy approximation and energy difference approximation to advanced regression and neural network machine learning algorithms employing complex descriptors of molecular compositions, geometries, and electronic structures are examined in conjunction with relevant literature examples. We discuss the interplay between ab initio data and machine learning (ML), i.e., whether it is better to base predictions on low-level quantum-chemical results corrected with ML or to bypass first-principles computations entirely and instead rely on elaborate deep learning architectures. Finally, we list currently available data sets of redox-active organic molecules and their experimental and/or computed properties to facilitate the development of screening platforms and rational design of redox-active organic molecules.

Machine Learning Methods for Small Data Challenges in Molecular Science

Small data are often used in scientific and engineering research due to the presence of various constraints, such as time, cost, ethics, privacy, security, and technical limitations in data acquisition. However, big data have been the focus for the past decade, small data and their challenges have received little attention, even though they are technically more severe in machine learning (ML) and deep learning (DL) studies. Overall, the small data challenge is often compounded by issues, such as data diversity, imputation, noise, imbalance, and high-dimensionality. Fortunately, the current big data era is characterized by technological breakthroughs in ML, DL, and artificial intelligence (AI), which enable data-driven scientific discovery, and many advanced ML and DL technologies developed for big data have inadvertently provided solutions for small data problems. As a result, significant progress has been made in ML and DL for small data challenges in the past decade. In this review, we summarize and analyze several emerging potential solutions to small data challenges in molecular science, including chemical and biological sciences. We review both basic machine learning algorithms, such as linear regression, logistic regression (LR), k-nearest neighbor (KNN), support vector machine (SVM), kernel learning (KL), random forest (RF), and gradient boosting trees (GBT), and more advanced techniques, including artificial neural network (ANN), convolutional neural network (CNN), U-Net, graph neural network (GNN), Generative Adversarial Network (GAN), long short-term memory (LSTM), autoencoder, transformer, transfer learning, active learning, graph-based semi-supervised learning, combining deep learning with traditional machine learning, and physical model-based data augmentation. We also briefly discuss the latest advances in these methods. Finally, we conclude the survey with a discussion of promising trends in small data challenges in molecular science.

Past, Present, and Future Perspectives on Computer-Aided Drug Design Methodologies

The application of computational approaches in drug discovery has been consolidated in the last decades. These families of techniques are usually grouped under the common name of "computer-aided drug design" (CADD), and they now constitute one of the pillars in the pharmaceutical discovery pipelines in many academic and industrial environments. Their implementation has been demonstrated to tremendously improve the speed of the early discovery steps, allowing for the proficient and rational choice of proper compounds for a desired therapeutic need among the extreme vastness of the drug-like chemical space. Moreover, the application of CADD approaches allows the rationalization of biochemical and interactive processes of pharmaceutical interest at the molecular level. Because of this, computational tools are now extensively used also in the field of rational 3D design and optimization of chemical entities starting from the structural information of the targets, which can be experimentally resolved or can also be obtained with other computer-based techniques. In this work, we revised the state-of-the-art computer-aided drug design methods, focusing on their application in different scenarios of pharmaceutical and biological interest, not only highlighting their great potential and their benefits, but also discussing their actual limitations and eventual weaknesses. This work can be considered a brief overview of computational methods for drug discovery.

Biological Activity Predictions of Ligands Based on Hybrid Molecular Fingerprinting and Ensemble Learning

The biological activity predictions of ligands are an important research direction, which can improve the efficiency and success probability of drug screening. However, the traditional prediction method has the disadvantages of complex modeling and low screening efficiency. Machine learning is considered an important research direction to solve these traditional method problems in the near future. This paper proposes a machine learning model with high predictive accuracy and stable prediction ability, namely, the back propagation neural network cross-support vector regression model (BPCSVR). By comparing multiple molecular descriptors, MACCS fingerprint and ECFP6 fingerprint were selected as inputs, and the stable prediction ability of the model was improved by integrating multiple models and correcting similar samples. We used leave-one-out cross-validation on 3038 samples from six data sets. The coefficient of determination, root mean square error, and absolute error were used as the evaluation parameters. After comparing the multiclass models, the results show that the BPCSVR model has stable prediction ability in different data sets, and the prediction accuracy is higher than other comparison models.

Guidance for good practice in the application of machine learning in development of toxicological quantitative structure-activity relationships (QSARs)

Recent years have seen a substantial growth in the adoption of machine learning approaches for the purposes of quantitative structure-activity relationship (QSAR) development. Such a trend has coincided with desire to see a shifting in the focus of methodology employed within chemical safety assessment: away from traditional reliance upon animal-intensive in vivo protocols, and towards increased application of in silico (or computational) predictive toxicology. With QSAR central amongst techniques applied in this area, the emergence of algorithms trained through machine learning with the objective of toxicity estimation has, quite naturally, arisen. On account of the pattern-recognition capabilities of the underlying methods, the statistical power of the ensuing models is potentially considerable-appropriate for the handling even of vast, heterogeneous datasets. However, such potency comes at a price: this manifesting as the general practical deficits observed with respect to the reproducibility, interpretability and generalisability of the resulting tools. Unsurprisingly, these elements have served to hinder broader uptake (most notably within a regulatory setting). Areas of uncertainty liable to accompany (and hence detract from applicability of) toxicological QSAR have previously been highlighted, accompanied by the forwarding of suggestions for "best practice" aimed at mitigation of their influence. However, the scope of such exercises has remained limited to "classical" QSAR-that conducted through use of linear regression and related techniques, with the adoption of comparatively few features or descriptors. Accordingly, the intention of this study has been to extend the remit of best practice guidance, so as to address concerns specific to employment of machine learning within the field. In doing so, the impact of strategies aimed at enhancing the transparency (feature importance, feature reduction), generalisability (cross-validation) and predictive power (hyperparameter optimisation) of algorithms, trained upon real toxicity data through six common learning approaches, is evaluated.