Search

Simplified molecular input line entry system (SMILES)-based deep learning models are slowly emerging as an important research topic in cheminformatics. In this study, we introduce SMILES pair encoding (SPE), a data-driven tokenization algorithm. SPE first learns a vocabulary of high-frequency SMILES substrings from a large chemical dataset (e.g., ChEMBL) and then tokenizes SMILES based on the learned vocabulary for the actual training of deep learning models. SPE augments the widely used atom-level tokenization by adding human-readable and chemically explainable SMILES substrings as tokens. Case studies show that SPE can achieve superior performances on both molecular generation and quantitative structure-activity relationship (QSAR) prediction tasks. In particular, the SPE-based generative models outperformed the atom-level tokenization model in the aspects of novelty, diversity, and ability to resemble the training set distribution. The performance of SPE-based QSAR prediction models were evaluated using 24 benchmark datasets where SPE consistently either did match or outperform atom-level and k-mer tokenization. Therefore, SPE could be a promising tokenization method for SMILES-based deep learning models. An open-source Python package SmilesPE was developed to implement this algorithm and is now freely available at https://github.com/XinhaoLi74/SmilesPE.

Histone deacetylase 3 (HDAC3) is a potential drug target for treatment of human diseases such as cancer, chronic inflammation, neurodegenerative diseases and diabetes. Machine learning (ML) as an essential cheminformatics approach has been widely used for QSAR modeling. However, none of them has been applied to HDAC3. To this end, we carefully compiled a set of 1098 compounds from the ChEMBL database that have been assayed against HDAC3 and calculated three different sets of molecular features for each compound, i. e. two-dimensional Mordred descriptors, MACCS keys (166 bits) and Morgan2 fingerprints (1024 bits). Five ML classifiers, i. e. k-Nearest Neighbour (KNN), Support Vector Machine (SVM), Random forest (RF), eXtreme Gradient Boosting (XGBoost) and Deep Neural Network (DNN) were trained on each feature set and optimized for classification. A total of 15 models were generated and carefully compared, among which the best-performing one was the XGBoost model based on the Morgan2 fingerprints, i. e. XGBoost_morgan2. Evaluated on a well-curated benchmarking set named MUBD-HDAC3, this model achieved a high early ROC enrichment (ROCE0.5 %: 41.02). A further retrospective screening of an annotated chemical library in PubChem demonstrated that the best model could identify 8 novel-scaffold HDAC3 inhibitors while assaying only 1 % of the compounds. To make this model accessible for the scientific community, we developed a python GUI application named HDAC3i-Finder to facilitate prospective screening for HDAC3 inhibitors. The source code of HDAC3i-Finder is available at https://github.com/jwxia2014/HDAC3i-Finder.

Deep neural networks can directly learn from chemical structures without extensive, user-driven selection of descriptors in order to predict molecular properties/activities with high reliability. But these approaches typically require large training sets to learn the endpoint-specific structural features and ensure reasonable prediction accuracy. Even though large datasets are becoming the new normal in drug discovery, especially when it comes to high-throughput screening or metabolomics datasets, one should also consider smaller datasets with challenging endpoints to model and forecast. Thus, it would be highly relevant to better utilize the tremendous compendium of unlabeled compounds from publicly-available datasets for improving the model performances for the user’s particular series of compounds. In this study, we propose the Molecular Prediction Model Fine-Tuning (MolPMoFiT) approach, an effective transfer learning method based on self-supervised pre-training + task-specific fine-tuning for QSPR/QSAR modeling. A large-scale molecular structure prediction model is pre-trained using one million unlabeled molecules from ChEMBL in a self-supervised learning manner, and can then be fine-tuned on various QSPR/QSAR tasks for smaller chemical datasets with specific endpoints. Herein, the method is evaluated on four benchmark datasets (lipophilicity, FreeSolv, HIV, and blood–brain barrier penetration). The results showed the method can achieve strong performances for all four datasets compared to other state-of-the-art machine learning modeling techniques reported in the literature so far.

Discovery of drugs against newly emerged pathogenic agents like the SARS-CoV-2 coronavirus (CoV) must be based on previous research against related species. Scientists need to get acquainted with and develop a global oversight over so-far tested molecules. Chemography (herein used Generative Topographic Mapping, in particular) places structures on a human-readable 2D map (obtained by dimensionality reduction of the chemical space of molecular descriptors) and is thus well suited for such an audit. The goal is to map medicinal chemistry efforts so far targeted against CoVs. This includes comparing libraries tested against various virus species/genera, predicting their polypharmacological profiles and highlighting often encountered chemotypes. Maps are challenged to provide predictive activity landscapes against viral proteins. Definition of “anti-CoV” map zones led to selection of therein residing 380 potential anti-CoV agents, out of a vast pool of 800M organic compounds.

Artificial intelligence (AI) is undergoing a revolution thanks to the breakthroughs of machine learning algorithms in computer vision, speech recognition, natural language processing and generative modelling. Recent works on publicly available pharmaceutical data showed that AI methods are highly promising for Drug Target prediction. However, the quality of public data might be different than that of industry data due to different labs reporting measurements, different measurement techniques, fewer samples and less diverse and specialized assays. As part of a European funded project (ExCAPE), that brought together expertise from pharmaceutical industry, machine learning, and high-performance computing, we investigated how well machine learning models obtained from public data can be transferred to internal pharmaceutical industry data. Our results show that machine learning models trained on public data can indeed maintain their predictive power to a large degree when applied to industry data. Moreover, we observed that deep learning derived machine learning models outperformed comparable models, which were trained by other machine learning algorithms, when applied to internal pharmaceutical company datasets. To our knowledge, this is the first large-scale study evaluating the potential of machine learning and especially deep learning directly at the level of industry-scale settings and moreover investigating the transferability of publicly learned target prediction models towards industrial bioactivity prediction pipelines.

Background: Molecular fingerprints are essential cheminformatics tools for virtual screening and mapping chemical space. Among the different types of fingerprints, substructure fingerprints perform best for small molecules such as drugs, while atom-pair fingerprints are preferable for large molecules such as peptides. However, no available fingerprint achieves good performance on both classes of molecules. Results: Here we set out to design a new fingerprint suitable for both small and large molecules by combining substructure and atom-pair concepts. Our quest resulted in a new fingerprint called MinHashed atom-pair fingerprint up to a diameter of four bonds (MAP4). In this fingerprint the circular substructures with radii of r = 1 and r = 2 bonds around each atom in an atom-pair are written as two pairs of SMILES, each pair being combined with the topological distance separating the two central atoms. These so-called atom-pair molecular shingles are hashed, and the resulting set of hashes is MinHashed to form the MAP4 fingerprint. MAP4 significantly outperforms all other fingerprints on an extended benchmark that combines the Riniker and Landrum small molecule benchmark with a peptide benchmark recovering BLAST analogs from either scrambled or point mutation analogs. MAP4 furthermore produces well-organized chemical space tree-maps (TMAPs) for databases as diverse as DrugBank, ChEMBL, SwissProt and the Human Metabolome Database (HMBD), and differentiates between all metabolites in HMBD, over 70 % of which are indistinguishable from their nearest neighbor using substructure fingerprints. Conclusion: MAP4 is a new molecular fingerprint suitable for drugs, biomolecules, and the metabolome and can be adopted as a universal fingerprint to describe and search chemical space. The source code is available at https://github.com/reymond-group/map4 and interactive MAP4 similarity search tools and TMAPs for various databases are accessible at http://map-search.gdb.tools/ and http://tm.gdb.tools/map4/.

We report a comprehensive fragment library with 205,903 fragments derived from the recently published Collection of Open Natural Products (COCONUT) data set with more than 400,000 non-redundant natural products. The natural products-based fragment library was compared with other two fragment libraries herein generated from ChEMBL (biologically relevant compounds) and Enamine-REAL (a large on-demand collection of synthetic compounds), both used as reference data sets with relevance in drug discovery. It was found that there is a large diversity of unique fragments derived from natural products and that the entire structures and fragments derived from natural products are more diverse and structurally complex than the two reference compound collections. During this work we introduced a novel visual representation of the chemical space based on the recently published concept of statistical-based database fingerprint. The compounds and fragments libraries from natural products generated and analyzed in this work are freely available.

The ccbmlib Python package is a collection of modules for modeling similarity value distributions based on Tanimoto coefficients for fingerprints available in RDKit. It can be used to assess the statistical significance of Tanimoto coefficients and evaluate how molecular similarity is reflected when different fingerprint representations are used. Significance measures derived from p-values allow a quantitative comparison of similarity scores obtained from different fingerprint representations that might have very different value ranges. Furthermore, the package models conditional distributions of similarity coefficients for a given reference compound. The conditional significance score estimates where a test compound would be ranked in a similarity search. The models are based on the statistical analysis of feature distributions and feature correlations of fingerprints of a reference database. The resulting models have been evaluated for 11 RDKit fingerprints, taking a collection of ChEMBL compounds as a reference data set. For most fingerprints, highly accurate models were obtained, with differences of 1% or less for Tanimoto coefficients indicating high similarity.

A great variety of computational approaches support drug design processes, helping in selection of new potentially active compounds, and optimization of their physicochemical and ADMET properties. Machine learning is a group of methods that are able to evaluate in relatively short time enormous amounts of data. However, the quality of machine-learning-based prediction depends on the data supplied for model training. In this study, we used deep neural networks for the task of compound activity prediction and developed dropout-based approaches for estimating prediction uncertainty. Several types of analyses were performed: the relationships between the prediction error, similarity to the training set, prediction uncertainty, number and standard deviation of activity values were examined. It was tested whether incorporation of information about prediction uncertainty influences compounds ranking based on predicted activity and prediction uncertainty was used to search for the potential errors in the ChEMBL database. The obtained outcome indicates that incorporation of information about uncertainty of compound activity prediction can be of great help during virtual screening experiments.

This work introduces a novel chemical space embedding method "Hilbert-Curve Assisted Structure Embedding (HCASE)" with help of pseudo-Hilbert Curves and Scaffold- Keys. The method was designed to produce an embedding that can be intuitively interpreted by medicinal chemists and data analysts. We analyzed the embedding of approved drug molecules (DrugBank) and natural products (CANVASS) into chemical spaces defined by Bemis-Murcko scaffolds extracted from ChEMBL (v24.1) database and from ChEMBL (v23) Natural Products. The implementation of HCASE algorithm and the input and results files of the analyses are available at https://github.com/ncats/hcase .

Background: The vast geographical expansion of novel coronavirus and an increasing number of COVID-19 affected cases have overwhelmed health and public health services. Artificial Intelligence (AI) and Machine Learning (ML) algorithms have extended their major role in tracking disease patterns, and in identifying possible treatments. Objective: This study aims to identify potential COVID-19 protease inhibitors through shape-based Machine Learning assisted by Molecular Docking and Molecular Dynamics simulations. Methods: 31 Repurposed compounds have been selected targeting the main coronavirus protease (6LU7) and a machine learning approach was employed to generate shape-based molecules starting from the 3D shape to the pharmacophoric features of their seed compound. Ligand-Receptor Docking was performed with Optimized Potential for Liquid Simulations (OPLS) algorithms to identify highaffinity compounds from the list of selected candidates for 6LU7, which were subjected to Molecular Dynamic Simulations followed by ADMET studies and other analyses. Results: Shape-based Machine learning reported remdesivir, valrubicin, aprepitant, and fulvestrant as the best therapeutic agents with the highest affinity for the target protein. Among the best shape-based compounds, a novel compound identified was not indexed in any chemical databases (PubChem, Zinc, or ChEMBL). Hence, the novel compound was named 'nCorv-EMBS'. Further, toxicity analysis showed nCorv-EMBS to be suitable for further consideration as the main protease inhibitor in COVID-19. Conclusion: Effective ACE-II, GAK, AAK1, and protease 3C blockers can serve as a novel therapeutic approach to block the binding and attachment of the main COVID-19 protease (PDB ID: 6LU7) to the host cell and thus inhibit the infection at AT2 receptors in the lung. The novel compound nCorv- EMBS herein proposed stands as a promising inhibitor to be evaluated further for COVID-19 treatment.

Background: The emergence of a new coronavirus (CoV), named 2019-nCoV, as an outbreak originated in the city of Wuhan, China, has resulted in the death of more than 3,400 people this year alone and has caused worldwide an alarming situation, particularly following previous CoV epidemics, including the Severe Acute Respiratory Syndrome (SARS) in 2003 and the Middle East Respiratory Syndrome (MERS) in 2012. Currently, no exists for infections caused by CoVs; however, some natural products may represent potential treatment resources, such as those that contain diterpenes. Objective: This study aimed to use computational methods to perform a virtual screening (VS) of candidate diterpenes with the potential to act as CoV inhibitors. Methods: 1,955 diterpenes, derived from the Nepetoideae subfamily (Lamiaceae), were selected using the SistematX tool (https://sistematx.ufpb.br), which were used to make predictions. From the ChEMBL database, 3 sets of chemical structures were selected for the construction of predictive models. Results: The chemical structures of molecules with known activity against SARS CoV, two of which were tested for activity against specific viral proteins and one of which was tested for activity against the virus itself, were classified according to their pIC50 values [-log IC50 (mol/l)]. Conclusion: In the consensus analysis approach, combining both ligand- and structure-based VSs, 19 compounds were selected as potential CoV inhibitors, including isotanshinone IIA (01), tanshinlactone (02), isocryptotanshinone (03), and tanshinketolactone (04), which did not present toxicity within the evaluated parameters.

Computational approaches to analyze various drug/ compound centered analysis often present a need to map attributes from multiple drug databases. In this study, we provide a Neo4j repository that integrates two of the most prominent open source drug databases, DrugBank and ChEMBL, with a goal of establishing an integrated data visualization and analysis tool for drug discovery studies. The drugs present in DrugBank are mapped to their counterparts in ChEMBL. The integration of these resources and the harmonization using knowledge graph serialization using Neo4j lead to identification of relationships between drugs and other related features that are otherwise spread across two different resources. A common data format, a prerequisite to populate the Neo4j database, enables users to identify new relationships central to drug discovery research, like Drug Target Interactions (DTI). The resource is freely available at: https://github.com/ambf0632/CompoundDB4j. © 2020 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

Ethnopharmacological relevance Yixin Ningshen tablet is a CFDA-approved TCM formula for treating depression clinically. However, little is known about its active compounds and related potential target proteins, so far, no researches have been performed to investigate its mechanism of action for the treatment of depression. Aim of the study Here we develop an original bioinformatics pipeline composed of text mining tools, database querying and systems biology combinatorial analysis, which is applied to rapidly explore the mechanism of action of Yixin Ningshen tablet in treating depression. Materials and methods Text mining and database query were applied to identify active compounds in Yixin Ningshen tablet for the treatment of depression. Then SwissTargetPrediction was used to predict their potential target proteins. PubMed was retrieved to summarize known depression related systems biology results. Ingenuity Pathway Analysis (IPA) tools and STRING were applied to construct a compound-target protein-gene protein-differential protein-differential metabolite network with the integration of compound-target interaction and systems biology results, as well as enrich the target proteins related pathways. ChEMBL and CDOCKER were used to validate the compound-target interactions. Results 62 active compounds and their 286 potential target proteins were identified in Yixin Ningshen tablet for the treatment of depression. The construction of compound-target protein-gene protein-differential protein-differential metabolite network shrinked the number of potential target proteins from 286 to 133. Pathway enrichment analysis of target proteins indicated that Neuroactive ligand-receptor interaction, Calcium signaling pathway, Serotonergic synapse, cAMP signaling pathway and Gap junction were the common primary pathways regulated by both Yixin Ningshen Tablet and anti-depressant drugs, and MAPK, Relaxin, AGE-RAGE, Estrogen, HIF-1, Jak-STAT signaling pathway, Endocrine resistance, Arachidonic acid metabolism and Regulation of actin cytoskeleton were the specifically main pathways regulated by Yixin Ningshen tablet for the treatment of depression. Further validations based on references and molecular docking results demonstrated that Yixin Ningshen tablet could primarily target MAPT, CHRM1 and DRD1, thus regulating serotonergic neurons, cholinergic transmission, norepinephrine and dopamine reuptake for the treatment of depression. Conclusions This study displays the power of extensive mining of public data and bioinformatical repositories to provide answers for a specific pharmacological question. It furthermore demonstrates how the usage of such a combinatorial approach is advantageous for the biologist in terms of experimentation time and costs.

Generative Topographic Mapping (GTM) can be efficiently used to visualize, analyze and model large chemical data. The GTM manifold needs to span the chemical space deemed relevant for a given problem. Therefore, the Frame set (FS) of compounds used for the manifold construction must well cover a given chemical space. Intuitively, the FS size must raise with the size and diversity of the target library. At the same time, the GTM training can be very slow or even becomes technically impossible at FS sizes of the order of 105 compounds – which is a very small number compared to today’s commercially accessible compounds, and, especially, to the theoretically feasible molecules. In order to solve this problem, we propose a Parallel GTM algorithm based on the merging of “intermediate” manifolds constructed in parallel for different subsets of molecules. An ensemble of these subsets forms a FS for the “final” manifold. In order to assess the efficiency of the new algorithm, 80 GTMs built on the FSs of different sizes ranging from 10 to 1.8M compounds selected from the ChEMBL database. Each GTM was challenged to build classification models for 712 biological activities (depending on the FS size). With the novel parallel GTM procedure, we could thus cover the entire spectrum of possible FS sizes, whereas previous studies were forced to rely on the working hypothesis that FS sizes of few thousands of compounds are sufficient to describe the ChEMBL chemical space. In fact, this study formally proves this to be true: a FS containing only 5000 randomly picked compounds is sufficient to represent the entire ChEMBL collection (1.8 M molecules), in the sense that a further increase of FS compound numbers has no benefice impact on the predictive propensity of the above-mentioned 712 activity classification models. Parallel GTM may, however, be required to generate maps based on very large FS, that might improve chemical space cartography of big commercial and virtual libraries, approaching billions of compounds.

Small molecule libraries for virtual screening are becoming a well-established tool for the identification of new hit compounds. As for experimental assays, the library quality, defined in terms of structural complexity and diversity, is crucial to increase the chance of a successful outcome in the screening campaign. In this context, Diversity-Oriented Synthesis has proven to be very effective, as the compounds generated are structurally complex and differ not only for the appendages, but also for the molecular scaffold. In this work, we automated the design of a library of lactams by applying a Diversity-Oriented Synthesis strategy called Build/Couple/Pair. We evaluated the novelty and diversity of these compounds by comparing them with lactam moieties contained in approved drugs, natural products, and bioactive compounds from ChEMBL. Finally, depending on their scaffold we classified them into β-, γ-, δ-, e-, and isolated, fused, bridged and spirolactam groups and we assessed their drug-like and lead-like properties, thus providing the value of this novel in silico designed library for medicinal chemistry applications.

Pharos is an integrated web-based informatics platform for the analysis of data aggregated by the Illuminating the Druggable Genome (IDG) Knowledge Management Center, an NIH Common Fund initiative. The current version of Pharos (as of October 2019) spans 20,244 proteins in the human proteome, 19,880 disease and phenotype associations, and 226,829 ChEMBL compounds. This resource not only collates and analyzes data from over 60 high-quality resources to generate these types, but also uses text indexing to find less apparent connections between targets, and has recently begun to collaborate with institutions that generate data and resources. Proteins are ranked according to a knowledge-based classification system, which can help researchers to identify less studied "dark" targets that could be potentially further illuminated. This is an important process for both drug discovery and target validation, as more knowledge can accelerate target identification, and previously understudied proteins can serve as novel targets in drug discovery. Two basic protocols illustrate the levels of detail available for targets and several methods of finding targets of interest. An Alternate Protocol illustrates the difference in available knowledge between less and more studied targets. © 2020 by John Wiley & Sons, Inc. Basic Protocol 1: Search for a target and view details Alternate Protocol: Search for dark target and view details Basic Protocol 2: Filter a target list to get refined results.

The objective of this work was to carry out a bibliographic survey of secondary metabolites isolated from the Velloziaceae family, creating a bank of compounds. After the bank was created, four prediction models for potentially active compounds against pathogenic microorganisms (Candida albicans, Escherichia coli, Pseudomonas aeruginosa and Salmonella sp.) were obtained trying to identify which metabolites would be more active against the strains. Four sets of compounds with known activity for microorganisms were selected for the construction of predictive models from the CHEMBL database. Another bank with 163 unique molecules isolated from the Velloziaceae family was built. The Volsurf+ v.1.0.7 software obtained the molecular descriptors and Knime 3.5 generated the in silico model. The performances of the internal and external tests were also analyzed. The study contributed through the virtual screening of a bank of metabolites to select several compounds with potential antimicrobial activity, highlighting the biflavonoid amentoflavone which showed potential activity against the four strains.

Beta-glucosidase inhibitors play important medical and biological roles. In this study, simple two-variable artificial neural network (ANN) classification models were developed for beta-glucosidase inhibitors screening. All bioassay data were obtained from the ChEMBL database. The classifiers were generated using 2D molecular descriptors and the data miner tool available in the STATISTICA package (STATISTICA Automated Neural Networks, SANN). In order to evaluate the models&rsquo, accuracy and select the best classifiers among automatically generated SANNs, the Matthews correlation coefficient (MCC) was used. The application of the combination of maxHBint3 and SpMax8_Bhs descriptors leads to the highest predicting abilities of SANNs, as evidenced by the averaged test set prediction results (MCC = 0.748) calculated for ten different dataset splits. Additionally, the models were analyzed employing receiver operating characteristics (ROC) and cumulative gain charts. The thirteen final classifiers obtained as a result of the model development procedure were applied for a natural compounds collection available in the BIOFACQUIM database. As a result of this beta-glucosidase inhibitors screening, eight compounds were univocally classified as active by all SANNs.

Nuclear receptors (NRs) are key regulators of human health and constitute a relevant target for medicinal chemistry applications as well as for toxicological risk assessment. Several open databases dedicated to small molecules that modulate NRs exist; however, depending on their final aim (i.e., adverse effect assessment or drug design), these databases contain a different amount and type of annotated molecules, along with a different distribution of experimental bioactivity values. Stemming from these considerations, in this work we aim to provide a unified dataset, NURA (NUclear Receptor Activity) dataset, collecting curated information on small molecules that modulate NRs, to be intended for both pharmacological and toxicological applications. NURA contains bioactivity annotations for 15,247 molecules and 11 selected NRs, and it was obtained by integrating and curating data from toxicological and pharmacological databases (i.e., Tox21, ChEMBL, NR-DBIND and BindingDB). Our results show that NURA dataset is a useful tool to bridge the gap between toxicology- and medicinal-chemistry-related databases, as it is enriched in terms of number of molecules, structural diversity and covered atomic scaffolds compared to the single sources. To the best of our knowledge, NURA dataset is the most exhaustive collection of small molecules annotated for their modulation of the chosen nuclear receptors. NURA dataset is intended to support decision-making in pharmacology and toxicology, as well as to contribute to data-driven applications, such as machine learning. The dataset and the data curation pipeline can be downloaded free of charge on Zenodo at the following DOI: https://doi.org/10.5281/zenodo.3991561.

Data sharing and reuse are crucial to enhance scientific progress and maximize return of investments in science. Although attitudes are increasingly favorable, data reuse remains difficult due to lack of infrastructures, standards, and policies. The FAIR (findable, accessible, interoperable, reusable) principles aim to provide recommendations to increase data reuse. Because of the broad interpretation of the FAIR principles, maturity indicators are necessary to determine the FAIRness of a dataset. In this work, we propose a reproducible computational workflow to assess data FAIRness in the life sciences. Our implementation follows principles and guidelines recommended by the maturity indicator authoring group and integrates concepts from the literature. In addition, we propose a FAIR balloon plot to summarize and compare dataset FAIRness. We evaluated the feasibility of our method on three real use cases where researchers looked for six datasets to answer their scientific questions. We retrieved information from repositories (ArrayExpress, Gene Expression Omnibus, eNanoMapper, caNanoLab, NanoCommons and ChEMBL), a registry of repositories, and a searchable resource (Google Dataset Search) via application program interfaces (API) wherever possible. With our analysis, we found that the six datasets met the majority of the criteria defined by the maturity indicators, and we showed areas where improvements can easily be reached. We suggest that use of standard schema for metadata and the presence of specific attributes in registries of repositories could increase FAIRness of datasets.

In the era of big data one of the key challenges is the conversion of data into knowledge by organization and searching. This is also true in the chemistry field, where novel technologies such as DNA encoded libraries, peptide libraries and new in silico enumeration methods produce immense amounts of molecules and related data. Handling these extremely large sets of molecules is tremendously complex and requires compromises that often come at the expense of interpretability. In this chapter we introduce and discuss an alternative, novel approach called “chemical topic modeling” which has been adopted from the text-mining field. This probabilistic framework offers an intuitive and meaningful way to organize large data sets. On the ChEMBL database (v23), an extremely heterogonous set of more than 1.6 million molecules, the method has proven its efficacy and robustness: a 100-topic model provided interesting topics like “proteins”, “DNA” or “steroids”. These rather general, yet nonetheless sensible and humanly understandable topics can provide the basis for further investigation.

The design of a chemical entity that potently and selectively binds to a biological target of therapeutic relevance has dominated the scene of drug discovery so far. However, recent findings suggest that multitarget ligands may be endowed with superior efficacy and be less prone to drug resistance. The Protein Data Bank (PDB) provides experimentally validated structural information about targets and bound ligands. Therefore, it represents a valuable source of information to help identifying active sites, understanding pharmacophore requirements, designing novel ligands, and inferring structure-activity relationships. In this study, we performed a large-scale analysis of the PDB by integrating different ligand-based and structure-based approaches, with the aim of identifying promising target associations for polypharmacology based on reported crystal structure information. First, the 2D and 3D similarity profiles of the crystallographic ligands were evaluated using different ligand-based methods. Then, activity data of pairs of similar ligands binding to different targets were inspected by comparing structural information with bioactivity annotations reported in the ChEMBL, BindingDB, BindingMOAD, and PDBbind databases. Afterward, extensive docking screenings of ligands in the identified cross-targets were made in order to validate and refine the ligand-based results. Finally, the therapeutic relevance of the identified target combinations for polypharmacology was evaluated from comparison with information on therapeutic targets reported in the Therapeutic Target Database (TTD). The results led to the identification of several target associations with high therapeutic potential for polypharmacology.

[Abstract] Sarcomas are a group of malignant neoplasms of connective tissue with a different etiology than carcinomas. The efforts to discover new drugs with antisarcoma activity have generated large datasets of multiple preclinical assays with different experimental conditions. For instance, the ChEMBL database contains outcomes of 37,919 different antisarcoma assays with 34,955 different chemical compounds. Furthermore, the experimental conditions reported in this dataset include 157 types of biological activity parameters, 36 drug targets, 43 cell lines, and 17 assay organisms. Considering this information, we propose combining perturbation theory (PT) principles with machine learning (ML) to develop a PTML model to predict antisarcoma compounds. PTML models use one function of reference that measures the probability of a drug being active under certain conditions (protein, cell line, organism, etc.). In this paper, we used a linear discriminant analysis and neural network to train and compare PT and non-PT models. All the explored models have an accuracy of 89.19–95.25% for training and 89.22–95.46% in validation sets. PTML-based strategies have similar accuracy but generate simplest models. Therefore, they may become a versatile tool for predicting antisarcoma compounds. Ministerio de Economía y Competitividad; CTQ2016-74881-P Ministerio de Economía y Competitividad; UNLC08-1E-002 Ministerio de Economía y Competitividad; UNLC13-13-3503 Xunta de Galicia; ED431C 2018/49 Xunta de Galicia; ED431D 2017/16 Xunta de Galicia; ED431G/01 Xunta de Galicia; ED431D 2017/23 Gobierno Vasco; IT1045-16 Instituto de Salud Carlos III; PI17/01826

Drug-decorated nanoparticles (DDNPs) have important medical applications. The current work combined Perturbation Theory with Machine Learning and Information Fusion (PTMLIF). Thus, PTMLIF models were proposed to predict the probability of nanoparticle&ndash, compound/drug complexes having antimalarial activity (against Plasmodium). The aim is to save experimental resources and time by using a virtual screening for DDNPs. The raw data was obtained by the fusion of experimental data for nanoparticles with compound chemical assays from the ChEMBL database. The inputs for the eight Machine Learning classifiers were transformed features of drugs/compounds and nanoparticles as perturbations of molecular descriptors in specific experimental conditions (experiment-centered features). The resulting dataset contains 107 input features and 249,992 examples. The best classification model was provided by Random Forest, with 27 selected features of drugs/compounds and nanoparticles in all experimental conditions considered. The high performance of the model was demonstrated by the mean Area Under the Receiver Operating Characteristics (AUC) in a test subset with a value of 0.9921 &plusmn, 0.000244 (10-fold cross-validation). The results demonstrated the power of information fusion of the experimental-centered features of drugs/compounds and nanoparticles for the prediction of nanoparticle&ndash, compound antimalarial activity. The scripts and dataset for this project are available in the open GitHub repository.

Funder: FP7 People: Marie-Curie Actions; doi: http://dx.doi.org/10.13039/100011264; Grant(s): 238701, 238701, An affinity fingerprint is the vector consisting of compound’s affinity or potency against the reference panel of protein targets. Here, we present the QAFFP fingerprint, 440 elements long in silico QSAR-based affinity fingerprint, components of which are predicted by Random Forest regression models trained on bioactivity data from the ChEMBL database. Both real-valued (rv-QAFFP) and binary (b-QAFFP) versions of the QAFFP fingerprint were implemented and their performance in similarity searching, biological activity classification and scaffold hopping was assessed and compared to that of the 1024 bits long Morgan2 fingerprint (the RDKit implementation of the ECFP4 fingerprint). In both similarity searching and biological activity classification, the QAFFP fingerprint yields retrieval rates, measured by AUC (~ 0.65 and ~ 0.70 for similarity searching depending on data sets, and ~ 0.85 for classification) and EF5 (~ 4.67 and ~ 5.82 for similarity searching depending on data sets, and ~ 2.10 for classification), comparable to that of the Morgan2 fingerprint (similarity searching AUC of ~ 0.57 and ~ 0.66, and EF5 of ~ 4.09 and ~ 6.41, depending on data sets, classification AUC of ~ 0.87, and EF5 of ~ 2.16). However, the QAFFP fingerprint outperforms the Morgan2 fingerprint in scaffold hopping as it is able to retrieve 1146 out of existing 1749 scaffolds, while the Morgan2 fingerprint reveals only 864 scaffolds.

The COVID-19 pandemic has negatively affected human life globally. It has led to economic crises and health emergencies across the world, spreading rapidly among the human population and has caused many deaths. Currently, there are no treatments available for COVID-19 so there is an urgent need to develop therapeutic interventions that could be used against the novel coronavirus infection. In this research, we used computational drug design technologies to repurpose existing drugs as inhibitors of SARS-CoV-2 viral proteins. The Broad Institute’s Drug Repurposing Hub consists of in-development/approved drugs and was computationally screened to identify potential hits which could inhibit protein targets encoded by the SARS-CoV-2 genome. By virtually screening the Broad collection, using rationally designed pharmacophore features, we identified molecules which may be repurposed against viral nucleocapsid and non-structural proteins. The pharmacophore features were generated after careful visualisation of the interactions between co-crystalised ligands and the protein binding site. The ChEMBL database was used to determine the compound’s level of inhibition of SARS-CoV-2 and correlate the predicted viral protein target with whole virus in vitro data. The results from this study may help to accelerate drug development against COVID-19 and the hit compounds should be progressed through further in vitro and in vivo studies on SARS-CoV-2., Graphical Abstract Communicated by Ramaswamy H. Sarma

DNA-Encoded Library (DEL) technology has emerged as an alternative method for bioactive molecules discovery in medicinal chemistry. It enables the simple synthesis and screening of compound libraries of enormous size. Even though it gains more and more popularity each day, there are almost no reports of chemoinformatics analysis of DEL chemical space. Therefore, in this project, we aimed to generate and analyze the ultra-large chemical space of DEL. Around 2500 DELs were designed using commercially available building blocks resulting in 2,5B DEL compounds that were compared to biologically relevant compounds from ChEMBL using Generative Topographic Mapping. This allowed to choose several optimal DELs covering the chemical space of ChEMBL to the highest extent and thus containing the maximum possible percentage of biologically relevant chemotypes. Different combinations of DELs were also analyzed to identify a set of mutually complementary libraries allowing to attain even higher coverage of ChEMBL than it is possible with one single DEL.

Much interest has been paid in the last decade on molecular predictors of promiscuity, including molecular weight, log P, molecular complexity, acidity constant and molecular topology, with correlations between promiscuity and those descriptors seemingly being context-dependent. It has been observed that certain therapeutic categories (e.g. mood disorders therapies) display a tendency to include multi-target agents (i.e. selective non-selectivity). Numerous QSAR models based on topological descriptors suggest that the topology of a given drug could be used to infer its therapeutic applications. Here, we have used descriptive statistics to explore the distribution of molecular topology descriptors and other promiscuity predictors across different therapeutic categories. Working with the publicly available ChEMBL database and 14 molecular descriptors, both hierarchical and non-hierchical clustering methods were applied to the descriptors mean values of the therapeutic categories after the refinement of the database (770 drugs grouped into 34 therapeutic categories). On the other hand, another publicly available database (repoDB) was used to retrieve cases of clinically-approved drug repositioning examples that could be classified into the therapeutic categories considered by the aforementioned clusters (111 cases), and the correspondence between the two studies was evaluated. Interestingly, a 3- cluster hierarchical clustering scheme based on only 14 molecular descriptors linked to promiscuity seem to explain up to 82.9% of approved cases of drug repurposing retrieved of repoDB. Therapeutic categories seem to display distinctive molecular patterns, which could be used as a basis for drug screening and drug design campaigns, and to unveil drug repurposing opportunities between particular therapeutic categories.

Optimization in medicinal chemistry often involves designing replacements for a section of a molecule which aim to retain potency while improving other properties of the compound. In this study, we perform a retrospective analysis using a number of computational methods to identify active side chains amongst a pool of random decoy side chains, mimicking a similar procedure that might be undertaken in a real medicinal chemistry project. We constructed a dataset derived from public ChEMBL and PDB data by identifying all ChEMBL assays where at least one of the compounds tested has also been co-crystallized in the PDB. Additionally, we required that there be at least ten active compounds tested in the same ChEMBL assay that are matched molecular pairs to the crystallized ligand. Using the compiled dataset consisting of sets of compounds from 402 assays, we have tested a number of methods for scoring side chains including Spark, a bioisostere replacement tool from Cresset, molecular docking using Glide from Schrodinger, docking with Smina, as well as other methods. In this work, we present a comparison of the performance of these methods in discriminating active side chains from decoys as well as recommendations for circumstances when different methods should be used.

We have developed a neural network potential energy function for use in drug discovery, with chemical element support extended from 41% to 94% of druglike molecules based on ChEMBL. We expand on the work of Smith et al., with their highly accurate network for the elements H, C, N, O, creating a network for H, C, N, O, S, F, Cl, P. We focus particularly on the calculation of relative conformer energies, for which we show that our new potential energy function has an RMSE of 0.70 kcal/mol for prospective druglike molecule conformers, substantially better than the previous state of the art. The speed and accuracy of this model could greatly accelerate the parameterization of protein-ligand binding free energy calculations for novel druglike molecules., 20 pages, 6 figures

Classical molecular dynamics (MD) techniques offer atomic-level insights into a wide range of physical systems, including biomolecular systems. They rely on empirical, parametrised equations (force fields) to describe the interaction potential between the constituents of the systems. Their predictive ability is critically dependent on the accuracy of these parameters. Highly optimised, well-validated parameters have been developed for many important biomolecules such as proteins, lipids and sugars. However, developing parameters for small molecules such as drugs is very challenging given the scale of chemical space. For instance, the ChEMBL database contains in excess of 1.7 million bioactive compounds. Although stand-alone software (e.g antechamber1,2) or web-servers (e.g. the Automated Topology Builder,3-5 atb.uq.edu.au) have been developed to assign parameters to drug-like molecules, they usually rely on fitting to quantum-mechanical (QM) calculations in combination with sets of empirical rules, and their applicability is limited to small molecules (less than a few tens of atoms). Automated parametrisation of large, bio-active, and often biologically relevant molecules is still impractical and greatly limits the predictive power of simulations involving such compounds.This thesis focuses on the use of graph-based approaches to develop new automated parametrisation paradigms which can exploit large data sets to simultaneously develop and assign force field parameters.An efficient Linear Programming method for deducing the charge state and bond order, based only on the molecular connectivity and chemical elements of its atoms, was developed. The algorithm was validated against the MMFF94 dataset containing 761 molecules with manually assigned bond orders. The approach was extended to i) cap molecular fragments (molecules having some atoms with incomplete valences) and ii) enumerate tautomeric (protomerism) forms of molecules. These extensions can be used for solving various problems related to drug-design and fragment-based empirical force-field parametrisation and can be applied to molecules containing hundreds of atoms.A new method for predicting chemically equivalent atoms in a molecule was developed. This method relies on identifying automorphisms of the molecule. False positives (non-chemically equivalent atoms treated as equivalent by the previous approach) were addressed by implementing a series of exceptions for double bonds, non-invertable rings and stereotopic atoms, respectively. This method was used to symmetrise force-field terms between chemically equivalent atoms in the Automated Topology Builder (ATB,3-5 atb.uq.edu.au).Building on these chemical equivalence relationships, a representation-independent, symmetry-corrected distance metric was developed (Blind RMSD). This facilitates the alignment of molecules with identical chemical graphs but different atom naming and indexing arbitrarily assigned. This approach was extended to fragments (i.e. subset of atoms in a molecules) for mapping graph nodes between two structures which may be equivalent in the 2D graph representation but not in a 3D model. How these procedures could be used to consolidate structural databases by identifying similar and differing conformational states is demonstrated.To address the problem of developing and assigning dihedral parameters in arbitrary molecules, a scheme for classifying dihedral terms around rotatable bonds based on the local substituents and their topologies (dihedral fragment) was developed. An analysis of the dihedral fragments contained in common biopolymers (DNA, RNA, proteins) and large chemical databases (PDB ligands, ChEMBL) revealed the extent of the diversity necessary to cover most of chemical space using this approach.A novel fitting technique (Fourier projections) that can fit scalar, 1-dimensional periodic functions up to an arbitrary precision using a series of sine and cosine functions with discrete (integer) frequencies was developed. I demonstrate how this approach can be used for the fitting of dihedral potentials based on quantum-mechanical (QM) data. The utility of the approach was demonstrated on the reparametrisation of the protein side chain dihedrals in the GROMOS54A7 force field, as well as on a large-scale automated reparametrisation of ∼1,200 dihedral fragments found in common drug-like molecules.Together, these tools and algorithms allowed the development of the Online tool for Fragment-based Molecule Parametrisation (OFraMP, oframp.atb.uq.edu.au). OFraMP is a web application that is designed to facilitate the parametrisation of atomic force fields for large (bio)molecules by matching sub-fragments within the target molecule to equivalent sub-fragments within a database of pre-existing parametrised molecules. Using two large and complex molecules (paclitaxel and a 290 atom-sized dendrimer) as examples, I illustrate how OFraMP can be used to semi-automatically assign Density Functional Theory (DFT)-derived point charges in a straight- forward and consistent manner.Many of the tools developed as part of this thesis are open-source and freely available to the scientific community under an MIT license.

Despite the achievements of antiretroviral therapy, discovery of new anti-HIV medicines remains an essential task because the existing drugs do not provide a complete cure for the infected patients, exhibit severe adverse effects, and lead to the appearance of resistant strains. To predict the interaction of drug-like compounds with multiple targets for HIV treatment, ligand-based drug design approach is widely applied. In this study, we evaluated the possibilities and limitations of (Q)SAR analysis aimed at the discovery of novel antiretroviral agents inhibiting the vital HIV enzymes. Local (Q)SAR models are based on the analysis of structure&ndash, activity relationships for molecules from the same chemical class, which significantly restrict their applicability domain. In contrast, global (Q)SAR models exploit data from heterogeneous sets of drug-like compounds, which allows their application to databases containing diverse structures. We compared the information for HIV-1 integrase, protease and reverse transcriptase inhibitors available in the EBI ChEMBL, NIAID HIV/OI/TB Therapeutics, and Clarivate Analytics Integrity databases as the sources for (Q)SAR training sets. Using the PASS and GUSAR software, we developed and validated a variety of (Q)SAR models, which can be further used for virtual screening of new antiretrovirals in the SAVI library. The developed models are implemented in the freely available web resource AntiHIV-Pred.

Retinoblastoma is a pediatric malignant tumor, common in children up to 5 years old. It is a disease commonly developed from retinoblasts, therefore the eye presents as the primary symptom leukocoria (white reflex that occurs when the retina is exposed) to this signal occurs due to displacement of the retina caused by tumor growth. The disease can affect only one eye (unilateral) or both eyes (bilateral). Depending on the region in which the tumor develops, the optic nerve and CNS may be compromised. Phenylpropanoids are widely studied anti-tumor bioactives in medicinal chemistry, which are present in several studies with good results against brain, breast, prostate and other tumors. This research aims to present, through in silico tools, bioactive with antitumor profile for retinoblastoma of 3,4,5-trihydroxycinnamic acid derivatives. For this, 128 derivatives were designed in ChemAxon © Marvin Sketch 18.21 program to obtain their 2D structures, then were imported into HyperChemTM 8.0.6 program (RMS 0.1kcal.mol-1 in 600 cycles) for the structures to have their structures. energies minimized by molecular mechanics (MM +) and semi-empirical method (AM1). A prediction model of biological activity was performed by the KNIME Analytics Platform 3.6, using molecules contained in the chemical structure database (ChEMBL, https://www.ebi.ac.uk/chembl) with known IC50 for proteins involved with the disease. Thus, other pharmaco-chemical parameters were also analyzed, such as ligand-receptor interaction through molecular docking, absorption rate, cytotoxicity risk prediction and CPCA chemometric analysis in the Volsurf + program to identify the molecular characteristics that best explain antitumor action. . We concluded that it was possible to elect 10 promising bioactive derivatives from the series worked in this research.

Introduction: Monoamine oxidase inhibitors (MAOIs) are compounds largely used in the treatment of Parkinson’s disease (PD), Alzheimer’s disease and other neuropsychiatric disorders since they are closely related to the MAO enzymes activity. The two isoforms of the MAO enzymes, MAO-A and MAO-B, are responsible for the degradation of monoamine neurotransmitters and due to this, relevant efforts have been devoted to finding new compounds with more selectivity and less side effects. One of the most used approaches is based on the use of computational approaches since they are time and money-saving and may allow us to find a more relevant structure-activity relationship. Objectives: In this manuscript, we will review the most relevant computational approaches aimed at the prediction and development of new MAO inhibitors. Subsequently, we will also introduce a new multitask model aimed at predicting MAO-A and MAO-B inhibitors. Methods: The QSAR multi-task model herein developed was based on the use of the linear discriminant analysis. This model was developed gathering 5,759 compounds from the public dataset Chembl. The molecular descriptors used was calculated using the Dragon software. Classical statistical tests were performed to check the validity and robustness of the model. Results: The herein proposed model is able to correctly classify all the 5,759 compounds. All the statistical performed tests indicated that this model is robust and reproducible. Conclusion: MAOIs are compounds of large interest since they are largely used in the treatment of very serious illness. These inhibitors may lose efficacy and produce severe side effects. Due to this, the development of selective MAO-A or MAO-B inhibitors is crucial for the treatment of these diseases and their effects. The herein proposed multi-target QSAR model may be a relevant tool in the development of new and more selective MAO inhibitors.

Drugs have become an essential part of our lives due to their ability to improve people's health and quality of life. However, for many diseases, approved drugs are not yet available or existing drugs have undesirable side effects, making the pharmaceutical industry strive to discover new drugs and active compounds. The development of drugs is an expensive process, which typically starts with the detection of candidate molecules (screening) after a protein target has been identified. To this end, the use of high-performance screening techniques has become a critical issue in order to palliate the high costs. Therefore, the popularity of computer-based screening (often called virtual screening or in silico screening) has rapidly increased during the last decade. A wide variety of Machine Learning (ML) techniques has been used in conjunction with chemical structure and physicochemical properties for screening purposes including (i) simple classifiers, (ii) ensemble methods, and more recently (iii) Multiple Classifier Systems (MCS). Here, we apply an MCS for virtual screening (D2-MCS) using circular fingerprints. We applied our technique to a dataset of cannabinoid CB2 ligands obtained from the ChEMBL database. The HTS collection of Enamine (1,834,362 compounds), was virtually screened to identify 48,232 potential active molecules using D2-MCS. Identified molecules were ranked to select 21 promising novel compounds for in vitro evaluation. Experimental validation confirmed six highly active hits (> 50% displacement at 10 mu M and subsequent Ki determination) and an additional five medium active hits (> 25% displacement at 10 mu M). Hence, D2-MCS provided a hit rate of 29% for highly active compounds and an overall hit rate of 52%. Xunta de Galicia. Consellería de Educación, Universidades e Formación Profesional Dutch Scientific Council Applied and Engineering Sciences (NWO-TTW)

Ethnopharmacological relevance Qishen Yiqi is a widely used Chinese herbal medicine formula with “qi invigorating and blood activating” property. Its dripping pill preparation (QSYQ) is a commercial herbal medicine approved by the China Food and Drug Administration (CFDA) in 2003 and is extensively used clinically to treat cardiovascular diseases, such as ischemic heart failure and angina pectoris, as well as for the secondary prevention of myocardial infarction. However, the bioactive ingredients of QSYQ remain unclear. As QSYQ is a compound herbal formula, it is of great importance to elucidate its pharmacologically active ingredients and underlying synergetic effects. Aim of the study This experimental study was conducted to comprehensively determine the combinatorial bioactive ingredients (CBIs) in QSYQ and to elucidate their potential synergetic effects. The established strategy may shed new light on how to rapidly determine CBIs in complex herbal formulas with holistic properties. Materials and methods An integrated evidence-based targeting strategy was introduced and validated to determine CBIs in QSYQ. The strategy included the following steps: (1) Chemical ingredients in QSYQ were analyzed via UPLC-Q-TOF/MS in the negative and positive modes and were identified by comparison with standard compounds and previously reported data. Their potential therapeutic activities were predicted based on the ChEMBL database to preliminarily search for candidate bioactive ingredients, and their combination was defined as the CBIs. (2) The CBIs were directly trapped and prepared from QSYQ with a two-dimensional chromatographic separation system, and the remaining part was defined as the rest ingredients (RIs). (3) As animal and cell models, left anterior descending coronary artery ligation (LAD)-induced heart failure in rats and hypoxia-induced cardiac myocyte injury in H9c2 cells were applied to compare the potency of QSYQ, CBIs and RIs. (4) The synergetic effects on cardiac myocyte protection of multiple ingredients in CBIs were examined in this cell model. Results (1) Forty-three ingredients in QSYQ were identified via UPLC-Q-TOF/MS. Based on evidence-based screening using the ChEMBL database, 24 ingredients were predicted to be bioactive ingredients, and their combination was considered the CBIs. (2) The CBIs and RIs were successfully prepared according to a two-dimensional chromatographic system. The CBIs were directly trapped and knocked out from QSYQ by hydrophilic interaction liquid chromatography coupled with reverse-phase liquid chromatography. The remaining part was used as RIs. (3) The results from pharmacological evaluation revealed that CBIs and QSYQ, but not RIs, significantly prevented myocardium injury; improved the ejection fraction (EF) and fractional shortening (FS); decreased the release of cardiac enzymes, including CK, CK-MB, and LDH; alleviated mitochondrial dysfunction; and protected the cell nucleus number and mitochondrial mass. Furthermore, QSYQ and CBIs possessed similar potency. (4) In hypoxia-stimulated H9c2 cells, CBIs showed far greater potency regarding the protection of cardiac myocyte injury than the individual ingredients in QSYQ, exhibiting obvious synergetic effects. Conclusions An integrated evidence-based targeting strategy was successfully established and validated to determine CBIs from QSYQ with excellent efficiency. Importantly, the holistic property of QSYQ was retained in the CBIs. Hence, this study may shed new light on how to rapidly reveal combinatorial bioactive ingredients from complex prescriptions and will be greatly helpful in the establishment of an appropriate approach to quality control for herbal medicines.

Unavailability of vaccines and effective drugs are primarily responsible for the growing menace of cryptosporidiosis. This study has incorporated a bioinformatics-based screening approach to explore potential vaccine candidates and novel drug targets inCryptosporidium parvumproteome. A systematic strategy was defined for comparative genomics, orthology with relatedCryptosporidiumspecies, prioritization parameters and MHC class I and II binding promiscuity. The approach reported cytoplasmic protein cgd7_1830, a signal peptide protein, as a novel drug target. SWISS-MODEL online server was used to generate the 3D model of the protein and was validated by PROCHECK. The model has been subjected toin silicodocking study with screened potent lead compounds from the ZINC database, PubChem and ChEMBL database using Flare software package of Cresset®. Furthermore, the approach reported protein cgd3_1400, as a vaccine candidate. The predicted B- and T-cell epitopes on the proposed vaccine candidate with highest scores were also subjected to docking study with MHC class I and II alleles using ClusPro web server. Results from this study could facilitate selection of proteins which could serve as drug targets and vaccine candidates to efficiently tackle the growing threat of cryptosporidiosis.

Identification of molecular targets and mechanism of action is always a challenge, in particular - for natural compounds due to inherent chemical complexity. BP-Cx-1 is a water-soluble modification of hydrolyzed lignin used as the platform for a portfolio of innovative pharmacological products aimed for therapy and supportive care of oncological patients. The present study describes a new approach, which combines in vitro screening of potential molecular targets for BP-Cx-1 using Diversity Profile - P9 panel by Eurofins Cerep (France) with a search of possible active components in silico in ChEMBL - manually curated chemical database of bioactive molecules with drug-like properties. The results of diversity assay demonstrate that BP-Cx-1 has multiple biological effects on neurotransmitters receptors, ligand-gated ion channels and transporters. Of particular importance is that the major part of identified molecular targets are involved in modulation of inflammation and immune response and might be related to tumorigenesis. Characterization of molecular composition of BP-Cx-1 with Fourier Transform Ion Cyclotron Resonance Mass Spectrometry and subsequent identification of possible active components by searching for molecular matches in silico in ChEMBL indicated polyphenolic components, nominally, flavonoids, sapogenins, phenanthrenes, as the major carriers of biological activity of BP-Cx-1. In vitro and in silico target screening yielded overlapping lists of proteins: adenosine receptors, dopamine receptor DRD4, glucocorticoid receptor, serotonin receptor 5-HT1, prostaglandin receptors, muscarinic cholinergic receptor, GABAA receptor. The pleiotropic molecular activities of polyphenolic components are beneficial in treatment of multifactorial disorders such as diseases associated with chronic inflammation and cancer.

A number of anti-retroviral drugs are being used for treating Human Immunodeficiency Virus (HIV) infection. Due to emergence of drug resistant strains, there is a constant quest to discover more effective anti-HIV compounds. In this endeavor, computational tools have proven useful in accelerating drug discovery. Although methods were published to design a class of compounds against a specific HIV protein, but an integrated web server for the same is lacking. Therefore, we have developed support vector machine based regression models using experimentally validated data from ChEMBL repository. Quantitative structure activity relationship based features were selected for predicting inhibition activity of a compound against HIV proteins namely protease (PR), reverse transcriptase (RT) and integrase (IN). The models presented a maximum Pearson correlation coefficient of 0.78, 0.76, 0.74 and 0.76, 0.68, 0.72 during tenfold cross-validation on IC50 and percent inhibition datasets of PR, RT, IN respectively. These models performed equally well on the independent datasets. Chemical space mapping, applicability domain analyses and other statistical tests further support robustness of the predictive models. Currently, we have identified a number of chemical descriptors that are imperative in predicting the compound inhibition potential. HIVprotI platform (http://bioinfo.imtech.res.in/manojk/hivproti) would be useful in virtual screening of inhibitors as well as designing of new molecules against the important HIV proteins for therapeutics development. Electronic supplementary material The online version of this article (10.1186/s13321-018-0266-y) contains supplementary material, which is available to authorized users.

Background Protein c-Myb is a therapeutic target. Some sesquiterpene lactones suppress Myb-dependent gene expression, which results in their potential anti-cancer activity. Material & methods Database ChEMBL is a representative of lactones for physicochemical and physiochemical properties. Data presented for 31 natural lactones are discussed in terms of quantitative structureactivity relationships with the objective to predict inhibitors of Myb-induced gene expression. Several constitutional descriptors are related to structure-activity. α-Methylene-γ-lactone groups enhance while OH functions worsen potency. The latter feature is in agreement with the fact that the more lipophilic the lactone, the greater the cytotoxicity because of the ability to cross lipoidal biomembranes. In general, numbers of π-systems and atoms, and polarizability enhance activity. Linear and nonlinear structure-activity models are developed, between lactones of a great structural diversity, to predict inhibitors of Myb-induced gene expression. Four variables (ML, UNC, TCO+OCOR, UNC+UNA) related to ATOM show a positive correlation because of the partial anionic and H-acceptor characters of O-atom. In most, CO group is conjugated. Result and conclusion Term OH shows negative coefficients because of the partial cationic quality of H-atom and because OH forms H-bonds with CO, causing them to be less H-acceptor. s-trans-s-trans-Germacranolide structure is the most active. Coefficients standard errors result acceptable in almost all equations. After cross-validation, linear equations for lactones, pseudoguaianolides and germacranolides are the most predictive. Most descriptors are constitutional variables.

Chagas disease is an endemic disease caused by Trypanosoma cruzi, which affects more than eight million people, mostly in the Americas. A search for new treatments is necessary to control and eliminate this disease. Sesquiterpene lactones (SLs) are an interesting group of secondary metabolites characteristic of the Asteraceae family that have presented a wide range of biological activities. From the ChEMBL database, we selected a diverse set of 4452, 1635, and 1322 structures with tested activity against the three T. cruzi parasitic forms: amastigote, trypomastigote, and epimastigote, respectively, to create random forest (RF) models with an accuracy of greater than 74 % for cross-validation and test sets. Afterward, a ligand-based virtual screen of the entire SLs of the Asteraceae database stored in SistematX (1306 structures) was performed. In addition, a structure-based virtual screen was also performed for the same set of SLs using molecular docking. Finally, using an approach combining ligand-based and structure-based virtual screening along with the equations proposed in this study to normalize the probability scores, we verified potentially active compounds and established a possible mechanism of action.

A deep learning (DL) method for quickly predicting surface charge density profiles (σ-profile) and cavity volumes (VCOSMO) of molecules for the COSMO-SAC model is developed. The molecular fingerprints are derived from the encoder state of a Transformer model pre-trained on the ChEMBL database, which allows transfer learning from large-scale unlabeled data and improve generalization performance by developing better molecular fingerprints for building models with significantly smaller datasets. Employing the pre-trained molecular fingerprints, a convolutional neural network (CNN) model for the σ-profile and VCOSMO prediction is trained and tested on the VT-2005 database. The obtained Transformer-CNN model presents superior performance to the GC-COSMO approach and enables the prediction of σ-profile and VCOSMO of millions of molecules in only a few minutes. Taking advantages of the model, a high-throughput solvent screening framework based on COSMO-SAC is further proposed and exemplified by searching sustainable solvent for the deterpenation process of citrus essential oils.

Combinational therapy is used for a long time in cancer treatment to overcome drug resistance related to monotherapy. Increased pharmacological data and the rapid development of deep learning methods have enabled the construction of models to predict and screen drug pairs. However, the size of drug libraries is restricted to hundreds to thousands of compounds. The ScaffComb framework, which aims to bridge the gaps in the virtual screening of drug combinations in large‐scale databases, is proposed here. Inspired by phenotype‐based drug design, ScaffComb integrates phenotypic information into molecular scaffolds, which can be used to screen the drug library and identify potent drug combinations. First, ScaffComb is validated using the US food and drug administration dataset and known drug combinations are successfully reidentified. Then, ScaffComb is applied to screen the ZINC and ChEMBL databases, which yield novel drug combinations and reveal an ability to discover new synergistic mechanisms. To our knowledge, ScaffComb is the first method to use phenotype‐based virtual screening of drug combinations in large‐scale chemical datasets., A phenotype‐based virtual screening pipeline of drug combinations in large‐scale chemical databases is proposed based on deep learning models. The phenotypic information is integrated into molecular scaffolds for drug screening. Screenings on the ZINC and ChEMBL databases demonstrate the potential of ScaffComb to identify potent drug combinations and novel synergistic mechanisms.

Health care systems have benefitted from rational drug discovery processes like vHTS, virtual high throughput screening pharmacophores and quantitative structure-activity relationships, and many challenges have been explored using such techniques: decisions on specificity and selectivity are made after screening millions of molecules for multiple targets. Recent challenges in drug research emphasize the design of drugs that bind with more than one target of interest (multi-target) and do not bind with undesirable targets. This work attempts to use a three-dimensional interaction profile of the active site of a class of proteins, identify selective positions for the binding of functional groups, called features, and develop ensembles of multi-targeted pharmacophores that retain specificity and selectivity. The goal of this study is to develop multi-target pharmacophores by computational methods using protein structures alone to guide the discovery of novel inhibitors of plasmepsins, displaying selectivity over their human homologs, cathepsin D and pepsin. The development of such novel tools is attempted using a combination of different approaches such as the molecular interaction field, clique graph and inductive logic programming to identify and compare specific and selective complementary features. The identification of selective combinations of features has resulted in the design of multi-featured specific and selective pharmacophores that are validated using antimalarial compounds in ChEMBL that are known for their anti-plasmepsin II activity. This novel method is computationally less intensive and is applicable to any known class of target structures for finding specific and selective binders simultaneously.

Introduction of a new drug to the market is a challenging and resource-consuming process. Predictive models developed with the use of artificial intelligence could be the solution to the growing need for an efficient tool which brings practical and knowledge benefits, but requires a large amount of high-quality data. The aim of our project was to develop quantitative structure–activity relationship (QSAR) model predicting serotonergic activity toward the 5-HT1A receptor on the basis of a created database. The dataset was obtained using ZINC and ChEMBL databases. It contained 9440 unique compounds, yielding the largest available database of 5-HT1A ligands with specified pKi value to date. Furthermore, the predictive model was developed using automated machine learning (AutoML) methods. According to the 10-fold cross-validation (10-CV) testing procedure, the root-mean-squared error (RMSE) was 0.5437, and the coefficient of determination (R2) was 0.74. Moreover, the Shapley Additive Explanations method (SHAP) was applied to assess a more in-depth understanding of the influence of variables on the model’s predictions. According to to the problem definition, the developed model can efficiently predict the affinity value for new molecules toward the 5-HT1A receptor on the basis of their structure encoded in the form of molecular descriptors. Usage of this model in screening processes can significantly improve the process of discovery of new drugs in the field of mental diseases and anticancer therapy.

Malaria is a life-threatening disease caused by Plasmodium parasites, which are transmitted to humans through infected female Anopheles mosquitoes. Over the past few decades, antimalarial resistance has posed one of the most important health challenges around the globe. To counteract the increased resistance to artemisinin combination therapy (ACT), the discovery of effective new antimalarial medicines is required at the moment. Plasmepsin X (PMX), an aspartic protease, is unravelled recently for playing an important role in egress and invasion of intraerythrocytic parasites by activating subtilisin-like protease 1. Therefore, the current study is centralized on PMX for in-silico identification of effective drugs. Structure based pharmacophore generation and virtual screening filtered out 47 candidates from almost 108 million entries catered in CHEMBL, PubChem, and ZINC databases. Five best hits were chosen based on docking scores. 1 µs molecular dynamics of homologue PMX and its complexes with the shortlisted five hit molecules and a known inhibitor 49c elucidated the stability of the complexes. Consistent with docking, the binding free energy estimated using MM-PBSA method recognized hits 1 (−144.4 kJ/mol) and 2 (−130.1 kJ/mol) the best inhibitors among all hits and the known inhibitor 49c (−68.7 kJ/mol). The new inhibitors can be put forward for further in-vivo studies as an antimalarial drug based on the comprehensive analyses reported herein.

ABCG2 is a substantial member of the ABC transporter superfamily that plays a significant role in multidrug resistance in cancer. Until recently, the 3D structure of ABCG2 has not been resolved, which resulted in the limitation of developing potential ABCG2 inhibitors using structure-based drug discovery. Herein, eMolecules, ChEMBL, and ChEBI databases, containing >25 million compounds, were virtually screened against the ABCG2 transporter in homodimer form. Performance of AutoDock4.2.6 software to predict inhibitor-ABCG2 binding mode and affinity were validated on the basis of available experimental data. The explored databases were filtered based on docking scores. The most potent hits with binding affinities higher than that of experimental bound ligand (MZ29) were then selected and subjected to molecular mechanics minimization, followed by binding energy calculation using molecular mechanics-generalized Born surface area (MM-GBSA) approach. Furthermore, molecular dynamics simulations for 50 ns, followed by MM-GBSA binding energy calculations, were performed for the promising compounds, unveiling eight potential inhibitors with binding affinities

Twenty eight new N2,N4-diphenylpyrimidine-2,4-diamines have been prepared in order to expand our understanding of the anti-malarial SAR of the scaffold. The aim of the study was to make structural modifications to improve the overall potency, selectivity and solubility of the series by varying the anilino groups attached to the 2- and 4-position. We evaluated the activity of the compounds against Plasmodium falciparum (Pf) 3D7, cytotoxicity against HepG2, % inhibition at a panel of 10 human kinases, solubility, permeability and lipophilicity, and human and rat in vitro clearance. 11 was identified as a potent anti-malarial with an IC50 of 0.66 µM at the 3D7 strain and a selectivity (SI) of ~ 40 in terms of cytotoxicity against the HepG2 cell line. It also displayed low experimental logD7.4 (2.27), reasonable solubility (124 µg/ml), good metabolic stability, but low permeability. A proteo-chemometric workflow was employed to identify putative Pf targets of the most promising compounds. Ligand-based similarity searching of the ChEMBL database led to the identification of most probable human targets. These were then used as input for sequence-based searching of the Pf proteome. Homology modelling and molecular docking were used to evaluate whether compounds could indeed bind to these targets with valid binding modes. In vitro biological testing against close human analogs of these targets was subsequently undertaken. This allowed us to identify potential Pf targets and human anti-targets that could be exploited in future development.

In the era of big data in Chemistry, the need for automated tools for virtual screening is compelling. Here, we present PrepFlow a toolkit for chemical library preparation and management. Starting from a list of compounds in SMILES or 2D molecular format, PrepFlow outputs a set of 3D molecular structures ready for use in subsequent drug discovery projects. Our development stands out for speed and robustness of execution, the efficient exploitation of HPC resources, and the implementation of an archiving strategy to save computer time, storage, and human intervention. Using a random selection of 600 compounds from available drug banks, we show that the preparation time per ligand on a desktop computer is 6.6 s. Thanks to these performances and the automatic parallelization on HPC, a chemical library of the size of ChEMBL (2 M) was prepared in around 3 days on a computer cluster. PrepFlow is freely distributed at the following link: https://ifm.chimie.unistra.fr/prepflow.