Compound Data Sets Research Articles

Investigasi Model Machine Learning Berbasis QSPR pada Inhibitor Korosi Pirimidin

Since corrosion causes considerable losses in many fields, including the economy, environment, society, industry, security, and safety, it is a major concern for the industrial and academic sectors. Damage control of materials based on organic compounds is currently a field of great interest. Because it is non-toxic, affordable, and effective in a variety of corrosive situations, pyrimidine has potential as a corrosion inhibitor. It takes a lot of time and resources to carry out experimental investigations in the exploration of potential corrosion inhibitor candidates. In this study, we evaluate the gradient boosting regressor (GBR), support vector regression (SVR), and k-nearest neighbor (KNN) algorithms as predictive models for corrosion inhibition efficiency using a machine learning (ML) approach based on the quantitative structure-property relationship model (QSPR). Based on the metric coefficient of determination (R2) and root mean square error (RMSE), we found that the GBR model had the best predictive performance compared to the SVR and KNN models as well as models from the literature for pyrimidine compound datasets. Overall, our study offers a new perspective on the ability of ML models to predict corrosion inhibition of iron surfaces

The combination of multi-approach studies to explore the potential therapeutic mechanisms of imidazole derivatives as an MCF-7 inhibitor in therapeutic strategies.

Breast cancer covers a large area of research because of its prevalence and high frequency all over the world. This study is based on drug discovery against breast cancer from a series of imidazole derivatives. A 3D-QSAR and activity atlas model was developed by exploring the dataset computationally, using the machine learning process of Flare. The dataset of compounds was divided into active and inactive compounds according to their biological and structural similarity with the reference drug. The obtained PLS regression model provided an acceptable r 2 = 0.81 and q2 = 0.51. Protein-ligand interactions of active molecules were shown by molecular docking against six potential targets, namely, TTK, HER2, GR, NUDT5, MTHFS, and NQO2. Then, toxicity risk parameters were evaluated for hit compounds. Finally, after all these screening processes, compound C10 was recognized as the best-hit compound. This study identified a new inhibitor C10 against cancer and provided evidence-based knowledge to discover more analogs.

Integrative transcriptomic and proteomic study of Zika viral infection reveals potential mechanisms for oncolytic therapy in neuroblastoma

Background: Paediatric neuroblastoma and brain tumours account for a third of all childhood cancer-related mortality. High-risk neuroblastoma is highly aggressive and survival is poor despite intensive multi-modal therapies with significant toxicity. Novel therapies are desperately needed. The Zika virus (ZIKV) is neurotropic and there is growing interest in employing ZIKV as a potential therapy against paediatric nervous system tumours, including neuroblastoma. Methods: Here, we perform extensive analysis of ZIKV infection studies to identify molecular mechanisms that may govern the oncolytic response in neuroblastoma cells. We summarise the neuroblastoma cell lines and ZIKV strains utilised and re-evaluate the infection data to deduce the susceptibility of neuroblastoma to the ZIKV oncolytic response. Integrating transcriptomics, interaction proteomics, dependency factor and compound datasets we show the involvement of multiple host systems during ZIKV infection. Results: We identified that most paediatric neuroblastoma cell lines are highly susceptible to ZIKV infection and that the PRVABC59 ZIKV strain is the most promising candidate for neuroblastoma oncolytic virotherapy. ZIKV induces TNF signalling, lipid metabolism, the Unfolded Protein Response (UPR), and downregulates cell cycle and DNA replication processes. ZIKV is dependent on sterol regulatory element binding protein (SREBP)-regulated lipid metabolism and three protein complexes; V-ATPase, ER Membrane Protein Complex (EMC) and mammalian translocon. We propose ZIKV non-structural protein 4B (NS4B) as a likely mediator of ZIKVs interaction with IRE1-mediated UPR, lipid metabolism and mammalian translocon. Conclusions: Our work provides a significant understanding of ZIKV infection in neuroblastoma cells, which will facilitate the progression of ZIKV-based oncolytic virotherapy through pre-clinical research and clinical trials.

Exploring activity landscapes with extended similarity: is Tanimoto enough?

Understanding structure-activity landscapes is essential in drug discovery. Similarly, it has been shown that the presence of activity cliffs in compound data sets can have a substantial impact not only on the design progress but also can influence the predictive ability of machine learning models. With the continued expansion of the chemical space and the currently available large and ultra-large libraries, it is imperative to implement efficient tools to analyze the activity landscape of compound data sets rapidly. The goal of this study is to show the applicability of the n-ary indices to quantify the structure-activity landscapes of large compound data sets using different types of structural representation rapidly and efficiently. We also discuss how a recently introduced medoid algorithm provides the foundation to finding optimum correlations between similarity measures and structure-activity rankings. The applicability of the n-ary indices and the medoid algorithm is shown by analyzing the activity landscape of 10 compound data sets with pharmaceutical relevance using three fingerprints of different designs, 16 extended similarity indices, and 11 coincidence thresholds.

Molecular Descriptors, Structure Generation, and Inverse QSAR/QSPR Based on SELFIES.

For inverse QSAR/QSPR in conventional molecular design, several chemical structures must be generated and their molecular descriptors must be calculated. However, there is no one-to-one correspondence between the generated chemical structures and molecular descriptors. In this paper, molecular descriptors, structure generation, and inverse QSAR/QSPR based on self-referencing embedded strings (SELFIES), a 100% robust molecular string representation, are proposed. A one-hot vector is converted from SELFIES to SELFIES descriptors x, and an inverse analysis of the QSAR/QSPR model y = f(x) with the objective variable y and molecular descriptor x is conducted. Thus, x values that achieve a target y value are obtained. Based on these values, SELFIES strings or molecules are generated, meaning that inverse QSAR/QSPR is performed successfully. The SELFIES descriptors and SELFIES-based structure generation are verified using datasets of actual compounds. The successful construction of SELFIES-descriptor-based QSAR/QSPR models with predictive abilities comparable to those of models based on other fingerprints is confirmed. A large number of molecules with one-to-one relationships with the values of the SELFIES descriptors are generated. Furthermore, as a case study of inverse QSAR/QSPR, molecules with target y values are generated successfully. The Python code for the proposed method is available at https://github.com/hkaneko1985/dcekit.

Using Machine Learning and Molecular Docking to Leverage Urease Inhibition Data for Virtual Screening.

Urease is a metalloenzyme that catalyzes the hydrolysis of urea, and its modulation has an important role in both the agricultural and medical industry. Even though numerous molecules have been tested against ureases of different species, their clinical translation has been limited due to chemical and metabolic stability as well as side effects. Therefore, screening new compounds against urease would be of interest in part due to rising concerns regarding antibiotic resistance. In this work, we collected and curated a diverse set of 2640 publicly available small-molecule inhibitors of jack bean urease and developed a classifier using a random forest machine learning method with high predictive performance. In addition, the physicochemical features of compounds were paired with molecular docking and protein-ligand fingerprint analysis to gather insight into the current activity landscape. We observed that the docking score could not differentiate active from inactive compounds within each chemical family, but scores were correlated with compound activity when all compounds were considered. Additionally, a decision tree model was built based on 2D and 3D Morgan fingerprints to mine patterns of the known active-class compounds. The final machine learning model showed good prediction performance against the test set (81% and 77% precision for active and inactive compounds, respectively). Finally, this model was employed, as a proof-of-concept, on an in-house library to predict new hits that were then tested against urease and found to be active. This is, to date, the largest, most diverse dataset of compounds used to develop predictive in silico models. Overall, the results highlight the usefulness of using machine learning classifiers and molecular docking to predict novel urease inhibitors.

Spectral density-based clustering algorithms for complex networks.

Clustering is usually the first exploratory analysis step in empirical data. When the data set comprises graphs, the most common approaches focus on clustering its vertices. In this work, we are interested in grouping networks with similar connectivity structures together instead of grouping vertices of the graph. We could apply this approach to functional brain networks (FBNs) for identifying subgroups of people presenting similar functional connectivity, such as studying a mental disorder. The main problem is that real-world networks present natural fluctuations, which we should consider. In this context, spectral density is an exciting feature because graphs generated by different models present distinct spectral densities, thus presenting different connectivity structures. We introduce two clustering methods: k-means for graphs of the same size and gCEM, a model-based approach for graphs of different sizes. We evaluated their performance in toy models. Finally, we applied them to FBNs of monkeys under anesthesia and a dataset of chemical compounds. We show that our methods work well in both toy models and real-world data. They present good results for clustering graphs presenting different connectivity structures even when they present the same number of edges, vertices, and degree of centrality. We recommend using k-means-based clustering for graphs when graphs present the same number of vertices and the gCEM method when graphs present a different number of vertices.

Molecular dynamics-derived pharmacophores of Schistosoma glutathione transferase in complex with bromosulfophthalein: Screening and analysis of potential inhibitors

Schistosoma glutathione transferases (GSTs) have been identified as attractive drug targets for the design of novel antischistosomals. Here, we used in silico methods to validate the discriminative inhibitory properties of bromosulfophthalein (BSP) against the 26-kDa GST from S. japonicum (Sj26GST), and the 28-kDa GST from S. haematobium (Sh28GST), versus human GST (hGST) isoforms alpha (hGSTA), mu (hGSTM) and pi (hGSTP). The use of BSP as an archetypal selective inhibitor was harnessed to produce molecular dynamics-derived pharmacophores of the two targets. Pharmacophore-based screening using a large dataset of experimental and approved drug compounds was performed to produce a shortlist of candidates. The top candidate for each target was prioritised via molecular docking, yielding guanosine-3′-monophosphate-5′-diphosphate (G3D) for Sj26GST, and quercetin-3′-O-phosphate (Q3P) for Sh28GST. Comparative molecular dynamics studies of both candidates compared to BSP showed similar characteristics of binding stability and strength, suggesting their potential to emulate the inhibitory effects of BSP.

Integrated use of ligand and structure-based virtual screening, molecular dynamics, free energy calculation and ADME prediction for the identification of potential PTP1B inhibitors.

Protein tyrosine phosphatases (PTPs) are the group of enzymes that control both cellular activity and the dephosphorylation of tyrosine (Tyr)-phosphorylated proteins. Dysregulation of PTP1B has contributed to numerous diseases including Diabetes Mellitus, Alzheimer's disease, and obesity rendering PTP1B as a legitimate target for therapeutic applications. It is highly challenging to target this enzyme because of its highly conserved and positively charged active-site pocket motivating researchers to find novel lead compounds against it. The present work makes use of an integrated approach combining ligand-based and structure-based virtual screening to find hit compounds targeting PTP1B. Initially, pharmacophore modeling was performed to find common features like two hydrogen bond acceptors, an aromatic ring and one hydrogen bond donor from the potent PTP1B inhibitors. The dataset of compounds matching with the common pharmacophoric features was filtered to remove Pan-Assay Interference substructure and to match the Lipinski criteria. Then, compounds were further prioritized using molecular docking and top fifty compounds with good binding affinity were selected for absorption, distribution, metabolism, and excretion (ADME) predictions. The top five compounds with high solubility, absorption and permeability holding score of - 10 to - 9.3kcal/mol along with Ertiprotafib were submitted to all-atom molecular dynamic (MD) studies. The MD studies and binding free energy calculations showed that compound M4, M5 and M8 were having better binding affinity for PTP1B enzyme with ∆Gtotal score of - 24.25, - 31.47 and - 33.81kcal/mol respectively than other compounds indicating that compound M8 could be a suitable lead compound as PTP1B inhibitor.

Structure-based virtual screening of novel natural products as chalcone derivatives against SARS-CoV-2 Mpro

Coronavirus disease 2019 (COVID-19), which is caused by SARS-CoV-2, has spread quickly around the world, causing a global pandemic. It has infected more than 500 million people as of April 28, 2022. Much research has been reported to stop the virus from spreading, but there are currently no approved medicines to treat COVID-19. In this work, a dataset of 142 natural products collected from various medicinal plants was used to perform structure-based virtual screening (SBVS) through the combined application of molecular docking and molecular dynamics (MD) simulation methods. First, the dataset of compounds was optimized using the density functional theory (DFT) approach. The optimized compounds were then submitted to the first screening, which was done by the pKCM web server to look for drug-likeness and the PyRx to look for binding affinity. Among the 142 natural substances, 10 compounds were selected for docking validation. Compounds that interact with CYS145 and LEU141, the essential catalytic residues, as well as compounds with binding affinities less than −8.0 kcal/mol, are considered promising anti-SARS-CoV-2 drug candidates. The top-ranked compounds were then evaluated by MD simulations and MM-GBSA method. These results could help researchers come up with new natural compounds that could be used to treat SARS-CoV-2. Communicated by Ramaswamy H. Sarma

Diagnosis of Multiple Faults in Rotating Machinery Using Ensemble Learning.

Fault diagnosis of rotating machines is an important task to prevent machinery downtime, and provide verifiable support for condition-based maintenance (CBM) decision-making. Deep learning-enabled fault diagnosis operations have become increasingly popular because features are extracted and selected automatically. However, it is challenging for these models to give superior results with rotating machine components of different scales, single and multiple faults across different rotating components, diverse operating speeds, and diverse load conditions. To address these challenges, this paper proposes a comprehensive learning approach with optimized signal processing transforms for single as well as multiple faults diagnosis across dissimilar rotating machine components: gearbox, bearing, and shaft. The optimized bicoherence, spectral kurtosis and cyclic spectral coherence feature spaces, and deep blending ensemble learning are explored for multiple faults diagnosis of these components. The performance analysis of the proposed approach has been demonstrated through a single joint training of the entire framework on a compound dataset containing multiple faults derived from three public repositories. A comparison with the state-of-the-art approaches that used these datasets, shows that our method gives improved results with different components and faults with nominal retraining.

Design, synthesis and biological characterization of novel activators of the TrkB neurotrophin receptor

Numerous studies have been published about the implication of the neurotrophin brain-derived neurotrophic factor (BDNF) and its receptor TrkB in the pathogenesis of several neurodegenerative conditions such as Alzheimer's disease, Parkinson's disease, Multiple Sclerosis and motor neuron disease. BDNF activates the TrkB receptor with high potency and specificity, promoting neuronal survival, differentiation and synaptic plasticity. Based on the main structural characteristics of LM22A-4, a previously published small molecule that acts as activator of the TrkB receptor, we have designed and synthesized a small data set of compounds. The lead idea for the design of the new compounds was to modify the third position of the LM22A-4, by introducing different substitutions in order to obtain compounds which will have not only better physicochemical properties but selective activity as well. ADME and toxicity profiles of molecules have been evaluated as well as their biological properties through the TrkB receptor and affinity to promote neurite differentiation.

Transfer Learning in Inorganic Compounds’ Crystal Structure Classification

Deep learning consists of deep convolutional layers and an unsupervised feature selection phase. The feature selection of deep learning on a large size dataset can be employed in correlated prediction models with small size datasets. This methodology is titled deep transfer learning model and enhances prediction model generalization. In this research, we proposed a prediction model for the crystal structure classification of inorganic compounds. Deep learning models in structure classification are usually trained using a large size dataset of 300 K compounds from different quantum compounds dataset (DS1). The feature selection of the deep learning models is reused for selecting features in a small size dataset (with 30 K inorganic compounds and containing 150 different crystal structures) and three alloy classes. The selected features are then fed into a random decision forest prediction model as input. The proposed convolutional neural network (CNN) with transfer learning realizes an accuracy of 98.5%. The experiment results display the CPU time consumed by our model, comparing the time required by similar models. The CPU classification time of the proposed model is 21 s on average.

Positive matrix factorization-based receptor modelling of particulate matter in northwest India

India is facing a dearth in the area of particulate matter (PM) source apportionment studies. Frequent exceedances and unusual high concentrations of PM observed in North India further presses the identification and quantification of particulate pollution sources to plan mitigation measures. But most of the PM source apportionment studies performed here are concentrated in and around Delhi, so this paper sets to explore the sources of PM in an understudied region of North India. The present study employed positive matrix factorization (PMF) 5.0 model developed by the United States Environmental Protection Agency (US EPA) for source characterisation of particulate matter quantified at Mohali in May 2012. Application of US EPA PMF 5.0 to a hybrid dataset of particulate matter and volatile organic compounds identified seven source factors, namingly, dust, wheat residue burning, biofuel use and waste disposal, mixed daytime, cars, two-wheelers) and industrial emissions and solvent use. The contributions of these individual source factors to PM10 and PM2.5 mass loadings were found to be 48.8%, 20.6%, 15.8%, 7.79%, 2.89%, 2.86%, 1.37% and 33.6%, 22.6%, 29.6%, 4.97%, 2.05%, 4.07% and 3.07% respectively. Dust and biomass burning were the major sources of PM in Mohali that contributed relatively highest average concentrations of PM10 (∼321 μg/m3) and PM2.5 (∼123 μg/m3) during morning hours (05:00–10:00 LT) in May 2012. This study reaffirmed the need to check anthropogenic activities and calls for strict air quality measures, especially in May to limit PM emissions in northwest India.

Enhancing developmental and reproductive toxicity knowledge: A new AOP stemming from glutathione depletion

Integrated approaches to testing and assessments (IATAs) have been proposed as a method to organise new approach methodologies in order to replace traditional animal testing for chemical safety assessments. To capture the mechanistic aspects of toxicity assessments, IATAs can be framed around the adverse outcome pathway (AOP) concept. To utilise AOPs fully in this context, a sufficient number of pathways need to be present to develop fit for purpose IATAs. In silico approaches can support IATA through the provision of predictive models and also through data integration to derive conclusions using a weight-of-evidence approach. To examine the maturity of a developmental and reproductive toxicity (DART) AOP network derived from the literature, an assessment of its coverage was performed against a novel toxicity dataset. A dataset of diverse compounds, with data from studies performed according to OECD test guidelines TG-421 and TG-422, was curated to test the performance of an in silico model based on the AOP network – allowing for the identification of knowledge gaps within the network. One such gap in the knowledge was filled through the development of an AOP stemming from the molecular initiating event ‘glutathione reaction with an electrophile’ leading to male fertility toxicity. The creation of the AOP provided the mechanistic rationale for the curation of pre-existing structural alerts to relevant key events. Integrating this new knowledge and associated alerts into the DART AOP network will improve its coverage of DART-relevant chemical space. In addition, broadening the coverage of AOPs for a particular regulatory endpoint may facilitate the development of, and confidence in, robust IATAs.

An Optimized Cloud Computing Method for Extracting Molecular Descriptors.

Extracting molecular descriptors from chemical compounds is an essential preprocessing phase for developing accurate classification models. Supervised machine learning algorithms offer the capability to detect "hidden" patterns that may exist in a large dataset of compounds, which are represented by their molecular descriptors. Assuming that molecules with similar structure tend to share similar physicochemical properties, large chemical libraries can be screened by applying similarity sourcing techniques in order to detect potential bioactive compounds against a molecular target. However, the process of generating these compound features is time-consuming. Our proposed methodology not only employs cloud computing to accelerate the process of extracting molecular descriptors but also introduces an optimized approach to utilize the computational resources in the most efficient way.

Generation of 2D-QSAR and pharmacophore models for fishing better anti-leishmanial therapeutics

Leishmaniasis, a life-threatening tropical disease that is endemic in nearly 100 countries, contributes to millions of deaths each year. However, very few antileishmanial compounds are available in the market and that too possess many drawbacks. Hence, the therapeutic arsenal requires potential and novel anti-leishmanial compounds to treat Leishmaniasis. In the present study, quantitative structure activity relationship (QSAR) model and Pharmacophore model were developed with a set of antileishmanial compounds collected from literature and commercial antileishmanial drugs. A ligand-based pharmacophore model was developed using active compound as template and it was used for searching the purchasable compound dataset of ZINC database for matching compounds. Thirteen novel, readily purchasable compounds were obtained from this approach, which shows good predicted activity, ADME and druglikeness. These compounds can be regarded as potential candidates to be developed as novel antileishmanial drugs with improved activity and reduced side effects.

Bimodal Emotion Recognition Based on Vocal and Facial Features

Emotion recognition is a crucial aspect of human communication, with applications in fields such as psychology, education, and healthcare. Identifying emotions accurately is challenging, as people use a variety of signals to express and perceive emotions. In this study, we address the problem of multimodal emotion recognition using both audio and video signals, to develop a robust and reliable system that can recognize emotions even when one modality is absent. To achieve this goal, we propose a novel architecture based on well-designed feature extractors for each modality and use model-level fusion based on a TFusion block to combine the information from both sources. To be more efficient in real-world scenarios, we trained our model on a compound dataset consisting of RAVDESS, RML, and eNTERFACE'05. It is then evaluated and compared to the state-of-the-art models. We find that our approach performs close to the modern solutions and can recognize emotions accurately when one of the modalities is missing. Additionally, we have developed a real-time emotion recognition application as a part of this work.

An algorithm to classify homologous series within compound datasets

Homologous series are groups of related compounds that share the same core structure attached to a motif that repeats to different degrees. Compounds forming homologous series are of interest in multiple domains, including natural products, environmental chemistry, and drug design. However, many homologous compounds remain unannotated as such in compound datasets, which poses obstacles to understanding chemical diversity and their analytical identification via database matching. To overcome these challenges, an algorithm to detect homologous series within compound datasets was developed and implemented using the RDKit. The algorithm takes a list of molecules as SMILES strings and a monomer (i.e., repeating unit) encoded as SMARTS as its main inputs. In an iterative process, substructure matching of repeating units, molecule fragmentation, and core detection lead to homologous series classification through grouping of identical cores. Three open compound datasets from environmental chemistry (NORMAN Suspect List Exchange, NORMAN-SLE), exposomics (PubChemLite for Exposomics), and natural products (the COlleCtion of Open NatUral producTs, COCONUT) were subject to homologous series classification using the algorithm. Over 2000, 12,000, and 5000 series with CH2 repeating units were classified in the NORMAN-SLE, PubChemLite, and COCONUT respectively. Validation of classified series was performed using published homologous series and structure categories, including a comparison with a similar existing method for categorising PFAS compounds. The OngLai algorithm and its implementation for classifying homologues are openly available at: https://github.com/adelenelai/onglai-classify-homologues.

Exploring quantitative structure-property relationship models for environmental fate assessment of petroleum hydrocarbons.

The rate and extent of biodegradation of petroleum hydrocarbons in the different aquatic environments is an important element to address. The major avenue for removing petroleum hydrocarbons from the environment is thought to be biodegradation. The present study involves the development of predictive quantitative structure-property relationship (QSPR) models for the primary biodegradation half-life of petroleum hydrocarbons that may be used to forecast the biodegradation half-life of untested petroleum hydrocarbons within the established models' applicability domain. These models use easily computable two-dimensional (2D) descriptors to investigate important structural characteristics needed for the biodegradation of petroleum hydrocarbons in freshwater (dataset 1), temperate seawater (dataset 2), and arctic seawater (dataset 3). All the developed models follow OECD guidelines. We have used double cross-validation, best subset selection, and partial least squares tools for model development. In addition, the small dataset modeler tool has been successfully used for the dataset with very few compounds (dataset 3 with 17 compounds), where dataset division was not possible. The resultant models are robust, predictive, and mechanisticallyinterpretable based on both internal and external validation metrics (R2 range of 0.605-0.959. Q2(Loo) range of 0.509-0.904, and Q2F1 range of 0.526-0.959). The intelligent consensus predictor tool has been used for theimprovement of the prediction quality for test set compounds which provided superior outcomes to those from individual partial least squares models based on several metrics (Q2F1 = 0.808 and Q2F2 = 0.805 for dataset 1 in freshwater). Molecular size and hydrophilic factor for freshwater, frequency of two carbon atoms at topological distance 4 for temperate seawater, and electronegative atom count relative to size for arctic seawater were found to be the most significant descriptors responsible for the regulation of biodegradation half-life of petroleum hydrocarbons.