Molecular similarity and diversity in chemoinformatics: from theory to applications

Design and Diversity Analysis of Chemical Libraries in Drug Discovery

Chemical libraries and compound data sets are among the main inputs to start the drug discovery process at universities, research institutes, and the pharmaceutical industry. The approach used in the design of compound libraries, the chemical information they possess, and the representation of structures, play a fundamental role in the development of studies: chemoinformatics, food informatics, in silico pharmacokinetics, computational toxicology, bioinformatics, and molecular modeling to generate computational hits that will continue the optimization process of drug candidates. The prospects for growth in drug discovery and development processes in chemical, biotechnological, and pharmaceutical companies began a few years ago by integrating computational tools with artificial intelligence methodologies. It is anticipated that it will increase the number of drugs approved by regulatory agencies shortly.

Chemical similarity of molecules with physiological response

Measuring the similarity among molecules is an important task in various chemically oriented problems. This elusive concept is hard to define and quantify. Moreover, the complexity of the problem is elevated by bifurcating the notion of molecular similarity to structural and chemical similarity. While the structural similarity of molecules is being extensively researched, the so-called chemical similarity is being mentioned scarcely. Here, we propose a way of converting the physico-chemical properties into molecular fingerprints. Then, using the apparatus of measuring the structural similarity, the chemical similarity can be assessed. The proof of a concept is demonstrated on a set of molecules that induce diverse physiological responses.

Network Science and Group Fusion Similarity-Based Searching to Explore the Chemical Space of Antiparasitic Peptides

Antimicrobial peptides (AMPs) have appeared as promising compounds to treat a wide range of diseases. Their clinical potentialities reside in the wide range of mechanisms they can use for both killing microbes and modulating immune responses. However, the hugeness of the AMPs' chemical space (AMPCS), represented by more than 10⁶⁵ unique sequences, has represented a big challenge for the discovery of new promising therapeutic peptides and for the identification of common structural motifs. Here, we introduce network science and a similarity searching approach to discover new promising AMPs, specifically antiparasitic peptides (APPs). We exploited the network-based representation of APPs' chemical space (APPCS) to retrieve valuable information by using three network types: chemical space (CSN), half-space proximal (HSPN), and metadata (METN). Some centrality measures were applied to identify in each network the most important and nonredundant peptides. Then, these central peptides were considered as queries (Qs) in group fusion similarity-based searches against a comprehensive collection of known AMPs, stored in the graph database StarPepDB, to propose new potential APPs. The performance of the resulting multiquery similarity-based search models (mQSSMs) was evaluated in five benchmarking data sets of APP/non-APPs. The predictions performed by the best mQSSM showed a strong-to-very-strong performance since their external Matthews correlation coefficient (MCC) values ranged from 0.834 to 0.965. Outstanding MCC values (>0.85) were attained by the mQSSM with 219 Qs from both networks CSN and HSPN with 0.5 as similarity threshold in external data sets. Then, the performance of our best mQSSM was compared with the APPs prediction servers AMPDiscover and AMPFun. The proposed model showed its relevance by outperforming state-of-the-art machine learning models to predict APPs. After applying the best mQSSM and additional filters on the non-APP space from StarPepDB, 95 AMPs were repurposed as potential APP hits. Due to the high sequence diversity of these peptides, different computational approaches were applied to identify relevant motifs for searching and designing new APPs. Lastly, we identified 11 promising APP lead candidates by using our best mQSSMs together with diversity-based network analyses, and 24 web servers for activity/toxicity and drug-like properties. These results support that network-based similarity searches can be an effective and reliable strategy to identify APPs. The proposed models and pipeline are freely available through the StarPep toolbox software at http://mobiosd-hub.com/starpep.

Computational discovery of novel anthelmintic natural compounds from Agave Brittoniana trel. Spp. Brachypus

Helminth infections are a medical problem in the world nowadays. This report used bond-based 2D quadratic indices, a bond-level QuBiLs-MAS molecular descriptor family, and Linear Discriminant Analysis (LDA) to obtain a quantitative linear model that discriminates between anthelmintic and non-anthelmintic drug-like organic-compounds. The model obtained correctly classified 87.46% and 81.82% of the training and external data sets, respectively. The developed model was used in a virtual screening to predict the biological activity of all chemicals (19) previously obtained and chemically characterized by some authors of this report from Agave brittoniana Trel. spp. Brachypus. The model identified several metabolites (12) as possible anthelmintics, and a group of 5 novel natural products was tested in an in vitro assay against Fasciola hepatica (100% effectivity at 500 µg/mL). Finally, the two best hits were evaluated in vivo in bald/c mice and the same helminth parasite using a 25 mg/kg dose. Compound 8 (Karatavinoside A) showed an efficacy of 92.2% in vivo. It is important to remark that this natural compound exhibits similar-to-superior activity as triclabendazole, the best human fasciolicide available in the market against Fasciola hepatica, resulting in a novel lead scaffold with anti-helminthic activity. Keywords: TOMOCOMD-CARDD Software; QuBiLs-MAS, nonstochastic and stochastic bond-based quadratic indices; LDA-based QSAR model; Computational Screening, Anthelmintic Agent; Agave brittoniana Trel. spp. Brachypus, Fasciola hepatica.

First report of q-RASAR modeling toward an approach of easy interpretability and efficient transferability

Quantitative structure-activity relationship (QSAR) and read-across techniques have recently been merged into a new emerging field of read-across structure-activity relationship (RASAR) that uses the chemical similarity concepts of read-across (an unsupervised step) and finally develops a supervised learning model (like QSAR). The RASAR method has so far been used only in case of graded predictions or classification modeling. In this work, we attempt, for the first time, to apply RASAR for quantitative predictions (q-RASAR) using a case study of androgen receptor binding affinity data. We have computed a number of error-based and similarity-based measures such as weighted standard deviation of the predicted values, coefficient of variation of the computed predictions, average similarity level of close training compounds for each query molecule, standard deviation and coefficient of variation of similarity levels, maximum similarity levels to positive and negative close training compounds, a concordance measure indicating similarity to positive, negative or both classes of close training compounds, etc. We have clubbed these additional measures along with the selected chemical descriptors from the previously developed QSAR model and redeveloped new partial least squares models from the training set, and predicted the endpoint using the query data set. Interestingly, these new models outperform the internal and external validation quality of the original QSAR model. In this study, we have also introduced a new similarity-based concordance measure (Banerjee-Roy coefficient) that can significantly contribute to the model quality. A q-RASAR model also has the advantage over read-across predictions in providing easy interpretation and indicating quantitative contributions of important chemical features. The strategy described here should be applicable to other biological/toxicological/property data modeling for enhanced quality of predictions, easy interpretability, and efficient transferability.

Leveraging algorithmic search in quantum chemical reaction path finding

Reaction path finding methods construct a graph connecting reactants and products in a quantum chemical energy landscape. They are useful in elucidating various reactions and provide footsteps for designing new reactions. Their enormous computational cost, however, limits their application to relatively simple reactions. This paper engages in accelerating reaction path finding by introducing the principles of algorithmic search. A new method called RRT/SC-AFIR is devised by combining rapidly exploring random tree (RRT) and single component artificial force induced reaction (SC-AFIR). Using 96 cores, our method succeeded in constructing a reaction graph for Fritsch-Buttenberg-Wiechell rearrangement within a time limit of 3 days, while the conventional methods could not. Our results illustrate that the algorithm theory provides refreshing and beneficial viewpoints on quantum chemical methodologies.

Selecting molecules with diverse structures and properties by maximizing submodular functions of descriptors learned with graph neural networks

Selecting diverse molecules from unexplored areas of chemical space is one of the most important tasks for discovering novel molecules and reactions. This paper proposes a new approach for selecting a subset of diverse molecules from a given molecular list by using two existing techniques studied in machine learning and mathematical optimization: graph neural networks (GNNs) for learning vector representation of molecules and a diverse-selection framework called submodular function maximization. Our method, called SubMo-GNN, first trains a GNN with property prediction tasks, and then the trained GNN transforms molecular graphs into molecular vectors, which capture both properties and structures of molecules. Finally, to obtain a subset of diverse molecules, we define a submodular function, which quantifies the diversity of molecular vectors, and find a subset of molecular vectors with a large submodular function value. This can be done efficiently by using the greedy algorithm, and the diversity of selected molecules measured by the submodular function value is mathematically guaranteed to be at least 63% of that of an optimal selection. We also introduce a new evaluation criterion to measure the diversity of selected molecules based on molecular properties. Computational experiments confirm that our SubMo-GNN successfully selects diverse molecules from the QM9 dataset regarding the property-based criterion, while performing comparably to existing methods regarding standard structure-based criteria. We also demonstrate that SubMo-GNN with a GNN trained on the QM9 dataset can select diverse molecules even from other MoleculeNet datasets whose domains are different from the QM9 dataset. The proposed method enables researchers to obtain diverse sets of molecules for discovering new molecules and novel chemical reactions, and the proposed diversity criterion is useful for discussing the diversity of molecular libraries from a new property-based perspective.

Comparisons of Molecular Structure Generation Methods Based on Fragment Assemblies and Genetic Graphs

The use of quantitative structure-property relationships (QSPRs) helps in predicting molecular properties for several decades, while the automatic design of new molecular structures is still emerging. The choice of algorithms to generate molecules is not obvious and is related to several factors such as the desired chemical diversity (according to an initial dataset's content) and the level of construction (the use of atoms, fragments, pattern-based methods). In this paper, we address the problem of molecular structure generation by revisiting two approaches: fragment-based methods (FMs) and genetic-based methods (GMs). We define a set of indices to compare generation methods on a specific task. New indices inform about the explored data space (coverage), compare how the data space is explored (representativeness), and quantifies the ratio of molecules satisfying requirements (generation specificity) without the use of a database composed of real chemicals as a reference. These indices were employed to compare generations of molecules fulfilling the desired property criterion, evaluated by QSPR.

Logistic matrix factorisation and generative adversarial neural network-based method for predicting drug-target interactions

Identifying drug-target protein association pairs is a prerequisite and a crucial task in drug discovery and development. Numerous computational models, based on different assumptions and algorithms, have been proposed as an alternative to the laborious, costly, and time-consuming traditional wet-lab methods. Most proposed methods focus on separated drug and target descriptors, calculated, respectively, from chemical structures and protein sequences, and fail to introduce and extract features where the interaction information is embedded. In this paper, we propose a new three-step method based on matrix factorisation and generative adversarial network (GAN) for drug-target interaction prediction. Firstly, the matrix factorisation technique is used to capture and extract the joint interaction feature, for both drugs and targets, from the drug-target interaction matrix. Then, a GAN is introduced for data augmentation. It generates a fake positive sample similar to the real positive sample (known interactions) in order to balance the samples, allow the exploitation of the entire negative sample, and increase the data size for an accurate prediction. Finally, a fully connected four-layer neural network is built for classification. Experimental results illustrate a higher prediction performance of the proposed method compared to shallow classifiers and to state-of-the-art methods with an accuracy higher than 97%. Moreover, the data generation effect is confirmed by evaluating the proposed method with and without the generation step. These results demonstrated the efficiency of the latent interaction features and data generation on predicting new drugs or repurposing existing drugs. Overview of the WGANMF-DTI workflow for the Drug-Target Interaction Prediction task.

Virtual screening of potential anticancer drugs based on microbial products

The development of microbial products for cancer treatment has been in the spotlight in recent years. In order to accelerate the lengthy and expensive drug development process, in silico screening tools are systematically employed, especially during the initial discovery phase. Moreover, considering the steadily increasing number of molecules approved by authorities for commercial use, there is a demand for faster methods to repurpose such drugs. Here we present a review on virtual screening web tools, such as publicly available databases of molecular targets and libraries of ligands, with the aim to facilitate the discovery of potential anticancer drugs based on microbial products. We provide an entry-level step-by-step description of the workflow for virtual screening of microbial metabolites with known protein targets, as well as two practical examples using freely available web tools. The first case presents a virtual screening study of drugs developed from microbial products using Caver Web, a web tool that performs docking along a tunnel. The second case comprises a comparative analysis between a wild type isocitrate dehydrogenase 1 and a mutant that results in cancer, using the recently developed web tool PredictSNP^Onco. In summary, this review provides the basic and essential background information necessary for virtual screening experiments, which may accelerate the discovery of novel anticancer drugs.

Attacking COVID-19 Progression Using Multi-Drug Therapy for Synergetic Target Engagement

COVID-19 is a devastating respiratory and inflammatory illness caused by a new coronavirus that is rapidly spreading throughout the human population. Over the past 12 months, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the virus responsible for COVID-19, has already infected over 160 million (>20% located in United States) and killed more than 3.3 million people around the world (>20% deaths in USA). As we face one of the most challenging times in our recent history, there is an urgent need to identify drug candidates that can attack SARS-CoV-2 on multiple fronts. We have therefore initiated a computational dynamics drug pipeline using molecular modeling, structure simulation, docking and machine learning models to predict the inhibitory activity of several million compounds against two essential SARS-CoV-2 viral proteins and their host protein interactors-S/Ace2, Tmprss2, Cathepsins L and K, and Mpro-to prevent binding, membrane fusion and replication of the virus, respectively. All together, we generated an ensemble of structural conformations that increase high-quality docking outcomes to screen over >6 million compounds including all FDA-approved drugs, drugs under clinical trial (>3000) and an additional >30 million selected chemotypes from fragment libraries. Our results yielded an initial set of 350 high-value compounds from both new and FDA-approved compounds that can now be tested experimentally in appropriate biological model systems. We anticipate that our results will initiate screening campaigns and accelerate the discovery of COVID-19 treatments.

Artificial intelligence and machine learning-aided drug discovery in central nervous system diseases: State-of-the-arts and future directions

Neurological disorders significantly outnumber diseases in other therapeutic areas. However, developing drugs for central nervous system (CNS) disorders remains the most challenging area in drug discovery, accompanied with the long timelines and high attrition rates. With the rapid growth of biomedical data enabled by advanced experimental technologies, artificial intelligence (AI) and machine learning (ML) have emerged as an indispensable tool to draw meaningful insights and improve decision making in drug discovery. Thanks to the advancements in AI and ML algorithms, now the AI/ML-driven solutions have an unprecedented potential to accelerate the process of CNS drug discovery with better success rate. In this review, we comprehensively summarize AI/ML-powered pharmaceutical discovery efforts and their implementations in the CNS area. After introducing the AI/ML models as well as the conceptualization and data preparation, we outline the applications of AI/ML technologies to several key procedures in drug discovery, including target identification, compound screening, hit/lead generation and optimization, drug response and synergy prediction, de novo drug design, and drug repurposing. We review the current state-of-the-art of AI/ML-guided CNS drug discovery, focusing on blood-brain barrier permeability prediction and implementation into therapeutic discovery for neurological diseases. Finally, we discuss the major challenges and limitations of current approaches and possible future directions that may provide resolutions to these difficulties.

LINGO-DL: a text-based approach for molecular similarity searching

The line notations of chemical structures are more compact than those of graphs and connection tables, so they can be useful for storing and transferring a large number of molecular structures. The simplified molecular input line system (SMILES) representation is the most extensively used, as it is much easier to utilise and comprehend than others, and it can be generated automatically from connection tables. A SMILES represents and encodes the molecule structure. It has been used by an existing method, LINGO, to calculate the molecular similarities and predict the structure-related properties. The LINGO method decomposes a canonical SMILES into a set of substrings of four characters referred to as LINGOs. The purpose of LINGO method is to measure the similarity between a pair of molecules by comparing the LINGOs that occur in each molecule. This paper aims to introduce an alternative version of the LINGO method using LINGOs of different lengths, called LINGO-DL. LINGO-DL is based on the fragmentation of canonical SMILES into substrings of three different lengths rather than one in LINGO method. Retrospective virtual screening experiments with MDDR, DUD, and MUV datasets show that the LINGO-DL outperforms the LINGO method, especially when the active molecules being sought have a high degree of structural heterogeneity.

Pathway-Based Drug-Repurposing Schemes in Cancer: The Role of Translational Bioinformatics

Cancer is a set of complex pathologies that has been recognized as a major public health problem worldwide for decades. A myriad of therapeutic strategies is indeed available. However, the wide variability in tumor physiology, response to therapy, added to multi-drug resistance poses enormous challenges in clinical oncology. The last years have witnessed a fast-paced development of novel experimental and translational approaches to therapeutics, that supplemented with computational and theoretical advances are opening promising avenues to cope with cancer defiances. At the core of these advances, there is a strong conceptual shift from gene-centric emphasis on driver mutations in specific oncogenes and tumor suppressors-let us call that the silver bullet approach to cancer therapeutics-to a systemic, semi-mechanistic approach based on pathway perturbations and global molecular and physiological regulatory patterns-we will call this the shrapnel approach. The silver bullet approach is still the best one to follow when clonal mutations in driver genes are present in the patient, and when there are targeted therapies to tackle those. Unfortunately, due to the heterogeneous nature of tumors this is not the common case. The wide molecular variability in the mutational level often is reduced to a much smaller set of pathway-based dysfunctions as evidenced by the well-known hallmarks of cancer. In such cases "shrapnel gunshots" may become more effective than "silver bullets". Here, we will briefly present both approaches and will abound on the discussion on the state of the art of pathway-based therapeutic designs from a translational bioinformatics and computational oncology perspective. Further development of these approaches depends on building collaborative, multidisciplinary teams to resort to the expertise of clinical oncologists, oncological surgeons, and molecular oncologists, but also of cancer cell biologists and pharmacologists, as well as bioinformaticians, computational biologists and data scientists. These teams will be capable of engaging on a cycle of analyzing high-throughput experiments, mining databases, researching on clinical data, validating the findings, and improving clinical outcomes for the benefits of the oncological patients.

Machine and deep learning approaches for cancer drug repurposing

Knowledge of the underpinnings of cancer initiation, progression and metastasis has increased exponentially in recent years. Advanced "omics" coupled with machine learning and artificial intelligence (deep learning) methods have helped elucidate targets and pathways critical to those processes that may be amenable to pharmacologic modulation. However, the current anti-cancer therapeutic armamentarium continues to lag behind. As the cost of developing a new drug remains prohibitively expensive, repurposing of existing approved and investigational drugs is sought after given known safety profiles and reduction in the cost barrier. Notably, successes in oncologic drug repurposing have been infrequent. Computational in-silico strategies have been developed to aid in modeling biological processes to find new disease-relevant targets and discovering novel drug-target and drug-phenotype associations. Machine and deep learning methods have especially enabled leaps in those successes. This review will discuss these methods as they pertain to cancer biology as well as immunomodulation for drug repurposing opportunities in oncologic diseases.

Genetic Algorithm-based Feature Selection Approach for Enhancing the Effectiveness of Similarity Searching in Ligand-based Virtual Screening

Background:

In the last years, similarity searching has gained wide popularity as a method for performing Ligand-Based Virtual Screening (LBVS). This screening technique functions by making a comparison of the target compound’s features with that of each compound in the database of compounds. It is well known that none of the individual similarity measures could provide the best performances each time pertaining to an active compound structure, representing all types of activity classes. In the literature, we find several techniques and strategies that have been proposed to improve the overall effectiveness of ligand-based virtual screening approaches.

Objective:

In this work, our main objective is to propose a features selection approach based on genetic algorithm (FSGASS) to improve similarity searching pertaining to ligand-based virtual screening.

Methods:

Our contribution allows us to identify the most important and relevant characteristics of chemical compounds and to minimize their number in their representations. This will allow the reduction of features space, the elimination of redundancy, the reduction of training execution time, and the increase of the performance of the screening process.

Results:

The obtained results demonstrate superiority in the performance compared with these obtained with Tanimoto coefficient, which is considered as the most widely coefficient to quantify the similarity in the domain of LBVS.

Conclusion:

Our results show that significant improvements can be obtained by using molecular similarity research methods at the basis of features selection.

Monomer structure fingerprints: an extension of the monomer composition version for peptide databases

Previously a fingerprint based on monomer composition (MCFP) of nonribosomal peptides (NRPs) has been introduced. MCFP is a novel method for obtaining a representative description of NRP structures from their monomer composition in a fingerprint form. An effective screening and prediction of biological activities has been obtained from Norine NRPs database. In this paper, we present an extension of the MCFP fingerprint. This extension is based on adding few columns into the fingerprint; representing monomer clusters, 2D structures, peptide categories, and peptide diversity. All these data have been extracted from the NRP structure. Experiments with Norine NRPs database showed that the extended MCFP, that can be called Monomer Structure FingerPrint (MSFP) produced high prediction accuracy (> 95%) together with a high recall rate (86%) obtained when MSFP was used for prediction and similarity searching. From this study it appeared that MSFP mainly built from monomer composition can substantially be improved by adding more columns representing useful information about monomer composition and 2D structure of NRPs.

Assessing the Performance of Mixed Strategies To Combine Lipophilic Molecular Similarity and Docking in Virtual Screening

The accuracy of structure-based (SB) virtual screening (VS) is heavily affected by the scoring function used to rank a library of screened compounds. Even in cases where the docked pose agrees with the experimental binding mode of the ligand, the limitations of current scoring functions may lead to sensible inaccuracies in the ability to discriminate between actives and inactives. In this context, the combination of SB and ligand-based (LB) molecular similarity may be a promising strategy to increase the hit rates in VS. This study explores different strategies that aim to exploit the synergy between LB and SB methods in order to mitigate the limitations of these techniques, and to enhance the performance of VS studies by means of a balanced combination between docking scores and three-dimensional (3D) similarity. Particularly, attention is focused to the use of measurements of molecular similarity with PharmScreen, which exploits the 3D distribution of atomic lipophilicity determined from quantum mechanical-based continuum solvation calculations performed with the MST model, in conjunction with three docking programs: Glide, rDock, and GOLD. Different strategies have been explored to combine the information provided by docking and similarity measurements for re-ranking the screened ligands. For a benchmarking of 44 datasets, including 41 targets, the hybrid methods increase the identification of active compounds, according to the early (ROCe%) and total (AUC) enrichment metrics of VS, compared to pure LB and SB methods. Finally, the hybrid approaches are also more effective in enhancing the chemical diversity of active compounds. The datasets employed in this work are available in https://github.com/Pharmacelera/Molecular-Similarity-and-Docking.

The efficiency of ligand-receptor interaction information alone as new descriptors in QSAR modeling via random forest artificial neural network

A new approach is introduced for the construction of a predictive quantitative structure-activity relationship model in which only ligand-receptor (LR) interaction features are used as relevant descriptors. This approach combines the benefit of the random forest (RF) as a new variable selection method with the intrinsic capability of the artificial neural network (ANN). The interaction information of the ligand-receptor (LR) complex was used as molecular docking descriptors. The most relevant descriptors were selected using the RF technique and used as inputs of ANN. The proposed RF ANN (RF-LM-ANN) method was optimized and then evaluated by the prediction of pEC₅₀ for some of the azine derivatives as non-nucleoside reverse transcriptase inhibitors. RF-LM-ANN model under the optimal conditions was evaluated using internal (validation) and external test sets. The determination coefficients of the external test and validation sets were 0.88 and 0.89, respectively. The mean square deviation (MSE) values for the prediction of biological activities in the external test and validation sets were found to be 0.10 and 0.11, respectively. The results obtained demonstrated the good prediction ability and high generalizability of the proposed RF-LM-ANN model based on the MMDs alone.

When global and local molecular descriptors are more than the sum of its parts: Simple, But Not Simpler?

In this report, we introduce a set of aggregation operators (AOs) to calculate global and local (group and atom type) molecular descriptors (MDs) as a generalization of the classical approach of molecular encoding using the sum of the atomic (or fragment) contributions. These AOs are implemented in a new and free software denominated MD-LOVIs ( http://tomocomd.com/md-lovis ), which allows for the calculation of MDs from atomic weights vector and LOVIs (local vertex invariants). This software was developed in Java programming language and employed the Chemical Development Kit (CDK) library for handling chemical structures and the calculation of atomic weights. An analysis of the complexities of the algorithms presented herein demonstrates that these aspects were efficiently implemented. The calculation speed experiments show that the MD-LOVIs software has satisfactory behavior when compared to software such as Padel, CDKDescriptor, DRAGON and Bluecal software. Shannon's entropy (SE)-based variability studies demonstrate that MD-LOVIs yields indices with greater information content when compared to those of popular academic and commercial software. A principal component analysis reveals that our approach captures chemical information orthogonal to that codified by the DRAGON, Padel and Mold2 software, as a result of the several generalizations in MD-LOVIs not used in other programs. Lastly, three QSARs were built using multiple linear regression with genetic algorithms, and the statistical parameters of these models demonstrate that the MD-LOVIs indices obtained with AOs yield better performance than those obtained when the summation operator is used exclusively. Moreover, it is also revealed that the MD-LOVIs indices yield models with comparable to superior performance when compared to other QSAR methodologies reported in the literature, despite their simplicity. The studies performed herein collectively demonstrated that MD-LOVIs software generates indices as simple as possible, but not simpler and that use of AOs enhances the diversity of the chemical information codified, which consequently improves the performance of traditional MDs.

Chemical space of naturally occurring compounds

The chemical space of naturally occurring compounds is vast and diverse. Other than biologics, naturally occurring small molecules include a large variety of compounds covering natural products from different sources such as plant, marine, and fungi, to name a few, and several food chemicals. The systematic exploration of the chemical space of naturally occurring compounds have significant implications in many areas of research including but not limited to drug discovery, nutrition, bio- and chemical diversity analysis. The exploration of the coverage and diversity of the chemical space of compound databases can be carried out in different ways. The approach will largely depend on the criteria to define the chemical space that is commonly selected based on the goals of the study. This chapter discusses major compound databases of natural products and cheminformatics strategies that have been used to characterize the chemical space of natural products. Recent exemplary studies of the chemical space of natural products from different sources and their relationships with other compounds are also discussed. We also present novel chemical descriptors and data mining approaches that are emerging to characterize the chemical space of naturally occurring compounds.

Natural Products in the "Marketplace": Interfacing Synthesis and Biology

Drugs are discovered through the biological screening of collections of compounds, followed by optimization toward functional end points. The properties of screening collections are often balanced between diversity, physicochemical favorability, intrinsic complexity, and synthetic tractability (Huggins, D. J.; et al. ACS Chem. Biol. 2011, 6, 208; DOI: 10.1021/cb100420r ). Whereas natural product (NP) collections excel in the first three attributes, NPs suffer a disadvantage on the last point. Academic total synthesis research has worked to solve this problem by devising syntheses of NP leads, diversifying late-stage intermediates, or derivatizing the NP target. This work has led to the discovery of reaction mechanisms, the invention of new methods, and the development of FDA-approved drugs. Few drugs, however, are themselves NPs; instead, NP analogues predominate. Here we highlight past examples of NP analogue development and successful NP-derived drugs. More recently, chemists have explored how NP analogues alter the retrosynthetic analysis of complex scaffolds, merging structural design and synthetic design. This strategy maintains the intrinsic complexity of the NP but can alter the physicochemical properties of the scaffold, like core instability that renders the NP a poor chemotype. Focused libraries based on these syntheses may exclude the NP but maintain the molecular properties that distinguish NP space from synthetic space (Stratton, C. F.; et al. Bioorg. Med. Chem. Lett. 2015, 25, 4802; DOI: 10.1016/j.bmcl.2015.07.014 ), properties that have statistical advantages in clinical progression (Luker, T.; et al. Bioorg. Med. Chem. Lett. 2011, 21, 5673, DOI: 10.1016/j.bmcl.2011.07.074 ; Ritchie, T. J.; Macdonald, S. J. F. Drug Discovery Today 2009, 14, 1011, DOI: 10.1016/j.drudis.2009.07.014 ). Research that expedites synthetic access to NP motifs can prevent homogeneity of chemical matter available for lead discovery. Easily accessed, focused libraries of NP scaffolds can fill empty but active gaps in screening sets and expand the molecular diversity of synthetic collections.

Machine learning in chemoinformatics and drug discovery

Chemoinformatics is an established discipline focusing on extracting, processing and extrapolating meaningful data from chemical structures. With the rapid explosion of chemical 'big' data from HTS and combinatorial synthesis, machine learning has become an indispensable tool for drug designers to mine chemical information from large compound databases to design drugs with important biological properties. To process the chemical data, we first reviewed multiple processing layers in the chemoinformatics pipeline followed by the introduction of commonly used machine learning models in drug discovery and QSAR analysis. Here, we present basic principles and recent case studies to demonstrate the utility of machine learning techniques in chemoinformatics analyses; and we discuss limitations and future directions to guide further development in this evolving field.

Spectrophores as one-dimensional descriptors calculated from three-dimensional atomic properties: applications ranging from scaffold hopping to multi-target virtual screening

Spectrophores are novel descriptors that are calculated from the three-dimensional atomic properties of molecules. In our current implementation, the atomic properties that were used to calculate spectrophores include atomic partial charges, atomic lipophilicity indices, atomic shape deviations and atomic softness properties. This approach can easily be widened to also include additional atomic properties. Our novel methodology finds its roots in the experimental affinity fingerprinting technology developed in the 1990’s by Terrapin Technologies. Here we have translated it into a purely virtual approach using artificial affinity cages and a simplified metric to calculate the interaction between these cages and the atomic properties. A typical spectrophore consists of a vector of 48 real numbers. This makes it highly suitable for the calculation of a wide range of similarity measures for use in virtual screening and for the investigation of quantitative structure–activity relationships in combination with advanced statistical approaches such as self-organizing maps, support vector machines and neural networks. In our present report we demonstrate the applicability of our novel methodology for scaffold hopping as well as virtual screening.

Finding Potential Multitarget Ligands Using PubChem

PubChem ( https://pubchem.ncbi.nlm.nih.gov ) is a key chemical information resource, developed and maintained by the US National Institutes of Health. The present chapter describes how to find potential multitarget ligands from PubChem that would be tested in further experiments. While the protocol presented here uses PubChem's Web-based interfaces to allow users to follow it interactively, it can also be implemented in computer software by using programmatic access interfaces to PubChem (such as PUG-REST or E-Utilities).

Analysing and Navigating Natural Products Space for Generating Small, Diverse, But Representative Chemical Libraries

Armed with the digital availability of two natural products libraries, amounting to some 195 885 molecular entities, we ask the question of how we can best sample from them to maximize their "representativeness" in smaller and more usable libraries of 96, 384, 1152, and 1920 molecules. The term "representativeness" is intended to include diversity, but for numerical reasons (and the likelihood of being able to perform a QSAR) it is necessary to focus on areas of chemical space that are more highly populated. Encoding chemical structures as fingerprints using the RDKit "patterned" algorithm, we first assess the granularity of the natural products space using a simple clustering algorithm, showing that there are major regions of "denseness" but also a great many very sparsely populated areas. We then apply a "hybrid" hierarchical K-means clustering algorithm to the data to produce more statistically robust clusters from which representative and appropriate numbers of samples may be chosen. There is necessarily again a trade-off between cluster size and cluster number, but within these constraints, libraries containing 384 or 1152 molecules can be found that come from clusters that represent some 18 and 30% of the whole chemical space, with cluster sizes of, respectively, 50 and 27 or above, just about sufficient to perform a QSAR. By using the online availability of molecules via the Molport system (www.molport.com), we are also able to construct (and, for the first time, provide the contents of) a small virtual library of available molecules that provided effective coverage of the chemical space described. Consistent with this, the average molecular similarities of the contents of the libraries developed is considerably smaller than is that of the original libraries. The suggested libraries may have use in molecular or phenotypic screening, including for determining possible transporter substrates.

Chemometric-assisted method development in hydrophilic interaction liquid chromatography: A review

With an enormous growth in the application of hydrophilic interaction liquid chromatography (HILIC), there has also been significant progress in HILIC method development. HILIC is a chromatographic method that utilises hydro-organic mobile phases with a high organic content, and a hydrophilic stationary phase. It has been applied predominantly in the determination of small polar compounds. Theoretical studies in computer-aided modelling tools, most importantly the predictive, quantitative structure retention relationship (QSRR) modelling methods, have attracted the attention of researchers and these approaches greatly assist the method development process. This review focuses on the application of computer-aided modelling tools in understanding the retention mechanism, the classification of HILIC stationary phases, prediction of retention times in HILIC systems, optimisation of chromatographic conditions, and description of the interaction effects of the chromatographic factors in HILIC separations. Additionally, what has been achieved in the potential application of QSRR methodology in combination with experimental design philosophy in the optimisation of chromatographic separation conditions in the HILIC method development process is communicated. Developing robust predictive QSRR models will undoubtedly facilitate more application of this chromatographic mode in a broader variety of research areas, significantly minimising cost and time of the experimental work.

ArbAlign: A Tool for Optimal Alignment of Arbitrarily Ordered Isomers Using the Kuhn-Munkres Algorithm

When assessing the similarity between two isomers whose atoms are ordered identically, one typically translates and rotates their Cartesian coordinates for best alignment and computes the pairwise root-mean-square distance (RMSD). However, if the atoms are ordered differently or the molecular axes are switched, it is necessary to find the best ordering of the atoms and check for optimal axes before calculating a meaningful pairwise RMSD. The factorial scaling of finding the best ordering by looking at all permutations is too expensive for any system with more than ten atoms. We report use of the Kuhn-Munkres matching algorithm to reduce the cost of finding the best ordering from factorial to polynomial scaling. That allows the application of this scheme to any arbitrary system efficiently. Its performance is demonstrated for a range of molecular clusters as well as rigid systems. The largely standalone tool is freely available for download and distribution under the GNU General Public License v3.0 (GNU_GPL_v3) agreement. An online implementation is also provided via a web server ( http://www.arbalign.org ) for convenient use.

Analysis of drug-endogenous human metabolite similarities in terms of their maximum common substructures

In previous work, we have assessed the structural similarities between marketed drugs (‘drugs’) and endogenous natural human metabolites (‘metabolites’ or ‘endogenites’), using ‘fingerprint’ methods in common use, and the Tanimoto and Tversky similarity metrics, finding that the fingerprint encoding used had a dramatic effect on the apparent similarities observed. By contrast, the maximal common substructure (MCS), when the means of determining it is fixed, is a means of determining similarities that is largely independent of the fingerprints, and also has a clear chemical meaning. We here explored the utility of the MCS and metrics derived therefrom. In many cases, a shared scaffold helps cluster drugs and endogenites, and gives insight into enzymes (in particular transporters) that they both share. Tanimoto and Tversky similarities based on the MCS tend to be smaller than those based on the MACCS fingerprint-type encoding, though the converse is also true for a significant fraction of the comparisons. While no single molecular descriptor can account for these differences, a machine learning-based analysis of the nature of the differences (MACCS_Tanimoto vs MCS_Tversky) shows that they are indeed deterministic, although the features that are used in the model to account for this vary greatly with each individual drug. The extent of its utility and interpretability vary with the drug of interest, implying that while MCS is neither ‘better’ nor ‘worse’ for every drug–endogenite comparison, it is sufficiently different to be of value. The overall conclusion is thus that the use of the MCS provides an additional and valuable strategy for understanding the structural basis for similarities between synthetic, marketed drugs and natural intermediary metabolites.

MetMaxStruct: A Tversky-Similarity-Based Strategy for Analysing the (Sub)Structural Similarities of Drugs and Endogenous Metabolites

BACKGROUND

Previous studies compared the molecular similarity of marketed drugs and endogenous human metabolites (endogenites), using a series of fingerprint-type encodings, variously ranked and clustered using the Tanimoto (Jaccard) similarity coefficient (TS). Because this gives equal weight to all parts of the encoding (thence to different substructures in the molecule) it may not be optimal, since in many cases not all parts of the molecule will bind to their macromolecular targets. Unsupervised methods cannot alone uncover this. We here explore the kinds of differences that may be observed when the TS is replaced-in a manner more equivalent to semi-supervised learning-by variants of the asymmetric Tversky (TV) similarity, that includes α and β parameters.

RESULTS

Dramatic differences are observed in (i) the drug-endogenite similarity heatmaps, (ii) the cumulative "greatest similarity" curves, and (iii) the fraction of drugs with a Tversky similarity to a metabolite exceeding a given value when the Tversky α and β parameters are varied from their Tanimoto values. The same is true when the sum of the α and β parameters is varied. A clear trend toward increased endogenite-likeness of marketed drugs is observed when α or β adopt values nearer the extremes of their range, and when their sum is smaller. The kinds of molecules exhibiting the greatest similarity to two interrogating drug molecules (chlorpromazine and clozapine) also vary in both nature and the values of their similarity as α and β are varied. The same is true for the converse, when drugs are interrogated with an endogenite. The fraction of drugs with a Tversky similarity to a molecule in a library exceeding a given value depends on the contents of that library, and α and β may be "tuned" accordingly, in a semi-supervised manner. At some values of α and β drug discovery library candidates or natural products can "look" much more like (i.e., have a numerical similarity much closer to) drugs than do even endogenites.

CONCLUSIONS

Overall, the Tversky similarity metrics provide a more useful range of examples of molecular similarity than does the simpler Tanimoto similarity, and help to draw attention to molecular similarities that would not be recognized if Tanimoto alone were used. Hence, the Tversky similarity metrics are likely to be of significant value in many general problems in cheminformatics.

Hybrid docking-QSAR studies of DPP-IV inhibition activities of a series of aminomethyl-piperidones

In this study, the dipeptidyl peptidase-IV (DPP-IV) inhibition activities of a series of novel aminomethyl-piperidones were investigated by molecular docking studies and modeled by quantitative structure-activity relationship (QSAR) methodology. Molecular docking studies were used to find the best conformations of the studied molecules in the active site of DPP-IV protein. Then the best docking-derived conformation for each molecule was applied for calculating the molecular descriptors. Multiple linear regression (MLR) and Levenberg-Marquardt artificial neural network (LM-ANN) were conducted on descriptors derived by docking. The results of these models revealed the superiority of LM-ANN model over MLR which showed the nonlinear relationship between the selected molecular descriptors and DPP-IV inhibition activities of studied molecules. The correlation coefficient (R) and standard error (SE) of ANN model were 0.983 and 0.103 for the training set and 0.966 and 0.168 for the external test set. These results showed a close agreement between the experimental and calculated values of pIC₅₀ which demonstrated the robustness of LM-ANN model in modeling of aminomethyl-piperidones. Applicability domain analysis and sensitivity analysis were applied on the obtained models. This study gives useful information for further experimental studies on DPP-IV inhibitors. The results of this work reveal the applicability of hybrid docking-QSAR methodology in ligand-protein studies.

Electrostatic Similarity Analysis of Human β-Defensin Binding in the Melanocortin System

The β-defensins are a class of small cationic proteins that serve as components of numerous systems in vertebrate biology, including the immune and melanocortin systems. Human β-defensin 3 (HBD3), which is produced in the skin, has been found to bind to melanocortin receptors 1 and 4 through complementary electrostatics, a unique mechanism of ligand-receptor interaction. This finding indicates that electrostatics alone, and not specific amino acid contact points, could be sufficient for function in this ligand-receptor system, and further suggests that other small peptide ligands could interact with these receptors in a similar fashion. Here, we conducted molecular-similarity analyses and functional studies of additional members of the human β-defensin family, examining their potential as ligands of melanocortin-1 receptor, through selection based on their electrostatic similarity to HBD3. Using Poisson-Boltzmann electrostatic calculations and molecular-similarity analysis, we identified members of the human β-defensin family that are both similar and dissimilar to HBD3 in terms of electrostatic potential. Synthesis and functional testing of a subset of these β-defensins showed that peptides with an HBD3-like electrostatic character bound to melanocortin receptors with high affinity, whereas those that were anticorrelated to HBD3 showed no binding affinity. These findings expand on the central role of electrostatics in the control of this ligand-receptor system and further demonstrate the utility of employing molecular-similarity analysis. Additionally, we identified several new potential ligands of melanocortin-1 receptor, which may have implications for our understanding of the role defensins play in melanocortin physiology.

Impact of similarity threshold on the topology of molecular similarity networks and clustering outcomes

BACKGROUND

Complex network theory based methods and the emergence of "Big Data" have reshaped the terrain of investigating structure-activity relationships of molecules. This change gave rise to new methods which need to face an important challenge, namely: how to restructure a large molecular dataset into a network that best serves the purpose of the subsequent analyses. With special focus on network clustering, our study addresses this open question by proposing a data transformation method and a clustering framework.

RESULTS

Using the WOMBAT and PubChem MLSMR datasets we investigated the relation between varying the similarity threshold applied on the similarity matrix and the average clustering coefficient of the emerging similarity-based networks. These similarity networks were then clustered with the InfoMap algorithm. We devised a systematic method to generate so-called "pseudo-reference" clustering datasets which compensate for the lack of large-scale reference datasets. With help from the clustering framework we were able to observe the effects of varying the similarity threshold and its consequence on the average clustering coefficient and the clustering performance.

CONCLUSIONS

We observed that the average clustering coefficient versus similarity threshold function can be characterized by the presence of a peak that covers a range of similarity threshold values. This peak is preceded by a steep decline in the number of edges of the similarity network. The maximum of this peak is well aligned with the best clustering outcome. Thus, if no reference set is available, choosing the similarity threshold associated with this peak would be a near-ideal setting for the subsequent network cluster analysis. The proposed method can be used as a general approach to determine the appropriate similarity threshold to generate the similarity network of large-scale molecular datasets.

Prioritization of anti-malarial hits from nature: chemo-informatic profiling of natural products with in vitro antiplasmodial activities and currently registered anti-malarial drugs

BACKGROUND

A large number of natural products have shown in vitro antiplasmodial activities. Early identification and prioritization of these natural products with potential for novel mechanism of action, desirable pharmacokinetics and likelihood for development into drugs is advantageous. Chemo-informatic profiling of these natural products were conducted and compared to currently registered anti-malarial drugs (CRAD).

METHODS

Natural products with in vitro antiplasmodial activities (NAA) were compiled from various sources. These natural products were sub-divided into four groups based on inhibitory concentration (IC50). Key molecular descriptors and physicochemical properties were computed for these compounds and analysis of variance used to assess statistical significance amongst the sets of compounds. Molecular similarity analysis, estimation of drug-likeness, in silico pharmacokinetic profiling, and exploration of structure-activity landscape were also carried out on these sets of compounds.

RESULTS

A total of 1040 natural products were selected and a total of 13 molecular descriptors were analysed. Significant differences were observed among the sub-groups of NAA and CRAD for at least 11 of the molecular descriptors, including number of hydrogen bond donors and acceptors, molecular weight, polar and hydrophobic surface areas, chiral centres, oxygen and nitrogen atoms, and shape index. The remaining molecular descriptors, including clogP, number of rotatable bonds and number of aromatic rings, did not show any significant difference when comparing the two compound sets. Molecular similarity and chemical space analysis identified natural products that were structurally diverse from CRAD. Prediction of the pharmacokinetic properties and drug-likeness of these natural products identified over 50% with desirable drug-like properties. Nearly 70% of all natural products were identified as potentially promiscuous compounds. Structure-activity landscape analysis highlighted compound pairs that form 'activity cliffs'. In all, prioritization strategies for the NAA were proposed.

CONCLUSIONS

Chemo-informatic profiling of NAA and CRAD have produced a wealth of information that may guide decisions and facilitate anti-malarial drug development from natural products. Articulation of the information provided within an interactive data-mining environment led to a prioritized list of NAA.

Identification of potential inhibitors based on compound proposal contest: Tyrosine-protein kinase Yes as a target

A search of broader range of chemical space is important for drug discovery. Different methods of computer-aided drug discovery (CADD) are known to propose compounds in different chemical spaces as hit molecules for the same target protein. This study aimed at using multiple CADD methods through open innovation to achieve a level of hit molecule diversity that is not achievable with any particular single method. We held a compound proposal contest, in which multiple research groups participated and predicted inhibitors of tyrosine-protein kinase Yes. This showed whether collective knowledge based on individual approaches helped to obtain hit compounds from a broad range of chemical space and whether the contest-based approach was effective.

The apparent permeabilities of Caco-2 cells to marketed drugs: magnitude, and independence from both biophysical properties and endogenite similarities

We bring together fifteen, nonredundant, tabulated collections (amounting to 696 separate measurements) of the apparent permeability (P app) of Caco-2 cells to marketed drugs. While in some cases there are some significant interlaboratory disparities, most are quite minor. Most drugs are not especially permeable through Caco-2 cells, with the median P app value being some 16 ⋅ 10(-6) cm s(-1). This value is considerably lower than those (1,310 and 230 ⋅ 10(-6) cm s(-1)) recently used in some simulations that purported to show that P app values were too great to be transporter-mediated only. While these values are outliers, all values, and especially the comparatively low values normally observed, are entirely consistent with transporter-only mediated uptake, with no need to invoke phospholipid bilayer diffusion. The apparent permeability of Caco-2 cells to marketed drugs is poorly correlated with either simple biophysical properties, the extent of molecular similarity to endogenous metabolites (endogenites), or any specific substructural properties. In particular, the octanol:water partition coefficient, logP, shows negligible correlation with Caco-2 permeability. The data are best explained on the basis that most drugs enter (and exit) Caco-2 cells via a multiplicity of transporters of comparatively weak specificity.

A Quantum-Based Similarity Method in Virtual Screening

One of the most widely-used techniques for ligand-based virtual screening is similarity searching. This study adopted the concepts of quantum mechanics to present as state-of-the-art similarity method of molecules inspired from quantum theory. The representation of molecular compounds in mathematical quantum space plays a vital role in the development of quantum-based similarity approach. One of the key concepts of quantum theory is the use of complex numbers. Hence, this study proposed three various techniques to embed and to re-represent the molecular compounds to correspond with complex numbers format. The quantum-based similarity method that developed in this study depending on complex pure Hilbert space of molecules called Standard Quantum-Based (SQB). The recall of retrieved active molecules were at top 1% and top 5%, and significant test is used to evaluate our proposed methods. The MDL drug data report (MDDR), maximum unbiased validation (MUV) and Directory of Useful Decoys (DUD) data sets were used for experiments and were represented by 2D fingerprints. Simulated virtual screening experiment show that the effectiveness of SQB method was significantly increased due to the role of representational power of molecular compounds in complex numbers forms compared to Tanimoto benchmark similarity measure.

Nonlinear QSAR modeling for predicting cytotoxicity of ionic liquids in leukemia rat cell line: an aid to green chemicals designing

Safety assessment and designing of safer ionic liquids (ILs) are among the priorities of the chemists and toxicologists today. Computational approaches have been considered as appropriate methods for prior safety assessment of chemicals and tools to aid in structural designing. The present study is an attempt to investigate the chemical attributes of a wide variety of ILs towards their cytotoxicity in leukemia rat cell line IPC-81 through the development of nonlinear quantitative structure-activity relationship (QSAR) models in the light of the OECD principles for QSAR development. Here, the cascade correlation network (CCN), probabilistic neural network (PNN), and generalized regression neural networks (GRNN) QSAR models were established for the discrimination of ILs in four categories of cytotoxicity and their end-point prediction using few simple descriptors. The diversity and nonlinearity of the considered dataset were evaluated through computing the Euclidean distance and Brock-Dechert-Scheinkman statistics. The constructed QSAR models were validated with external test data. The predictive power of these models was established through a variety of stringent parameters recommended in QSAR literature. The classification QSARs rendered the accuracy of >86%, and the regression models yielded correlation (R(2)) of >0.90 in test data. The developed QSAR models exhibited high statistical confidence and identified the structural elements of the ILs responsible for their cytotoxicity and, hence, could be useful tools in structural designing of safer and green ILs.

Virtual Screening for HIV Protease Inhibitors Using a Novel Database Filtering Procedure

A virtual screening to find novel inhibitors for HIV protease was performed on the ZINC database.1 A critical part in virtual screening and associated techniques is preliminary database filtering and size reduction and for that purpose a novel feature matrix matching procedure was used. The reduction of ∼14 million available ligands to a subset of 14299 ligands was achieved with a structure based approach where the analysis of the 3D structure of the active site of the protease produced a graph with hydrogen bond donor, hydrogen bond acceptor and hydrophobic subsites represented as graph nodes. A similar treatment was also applied to the compound database content and the comparison of binding site and ligand graphs was used to preselect potentially active ligands. The resulting set was further subjected to docking. The algorithm used was able to find several novel as well as previously known and experimentally tested ligands, demonstrating the validity of the approach.