Predicting an individual's cerebellar activity from functional connectivity fingerprints.

The cerebellum is gaining scientific attention as a key neural substrate of cognitive function; however, individual differences in the cerebellar organization have not yet been well studied. Individual differences in functional brain organization can be closely tied to individual differences in brain connectivity. 'Connectome Fingerprinting' is a modeling approach that predicts an individual's brain activity from their connectome. Here, we extend 'Connectome Fingerprinting' (CF) to the cerebellum. We examined functional MRI data from 160 subjects (98 females) of the Human Connectome Project young adult dataset. For each of seven cognitive task paradigms, we constructed CF models from task activation maps and resting-state cortico-cerebellar functional connectomes, using a set of training subjects. For each model, we then predicted task activation in novel individual subjects, using their resting-state functional connectomes. In each cognitive paradigm, the CF models predicted individual subject cerebellar activity patterns with significantly greater precision than did predictions from the group average task activation. Examination of the CF models revealed that the cortico-cerebellar connections that carried the most information were those made with the non-motor portions of the cerebral cortex. These results demonstrate that the fine-scale functional connectivity between the cerebral cortex and cerebellum carries important information about individual differences in cerebellar functional organization. Additionally, CF modeling may be useful in the examination of patients with cerebellar dysfunction, since model predictions require only resting-state fMRI data which is more easily obtained than task fMRI.

A consensual machine-learning-assisted QSAR model for effective bioactivity prediction of xanthine oxidase inhibitors using molecular fingerprints.

Xanthine oxidase inhibitors (XOIs) have been widely studied due to the promising potential as safe and effective therapeutics in hyperuricemia and gout. Currently, available XOI molecules have been developed from different experiments but they are with the wide structure diversity and significant varying bioactivities. So it is of great practical significance to present a consensual QSAR model for effective bioactivity prediction of XOIs based on a systematic compiling of these XOIs across different experiments. In this work, 249 XOIs belonging to 16 scaffolds were collected and were integrated into a consensual dataset by introducing the concept of IC50 values relative to allopurinol (RIC50). Here, extended connectivity fingerprints (ECFPs) were employed to represent XOI molecules. By performing effective feature selection by machine-learning method, 54 crucial fingerprints were indicated to be valuable for predicting the inhibitory potency (IP) of XOIs. The optimal predictor yields the promising performance by different cross-validation tests. Besides, an external validation of 43 XOIs and a case study on febuxostat also provide satisfactory results, indicating the powerful generalization of our predictor. Here, the predictor was interpreted by shapely additive explanation (SHAP) method which revealed several important substructures by mapping the featured fingerprints to molecular structures. Then, 15 new molecules were designed and predicted by our predictor to show superior IP than febuxostat. Finally, molecular docking simulation was performed to gain a deep insight into molecular binding mode with xanthine oxidase (XO) enzyme, showing that molecules with selenazole moiety, cyano group and isopropyl group tended to yield higher IP. The absorption, distribution, metabolism, excretion and toxicity (ADMET) prediction results further enhanced the potential of these novel XOIs as drug candidates. Overall, this work presents a QSAR model for accurate prediction of IP of XOIs, and is expected to provide new insights for further structure-guided design of novel XOIs.

Direct Prediction of Bioaccumulation of Organic Contaminants in Plant Roots from Soils with Machine Learning Models Based on Molecular Structures.

Root concentration factor (RCF) is an important characterization parameter to describe accumulation of organic contaminants in plants from soils in life cycle impact assessment (LCIA) and phytoremediation potential assessment. However, building robust predictive models remains challenging due to the complex interactions among chemical-soil-plant root systems. Here we developed end-to-end machine learning models to devolve the complex molecular structure relationship with RCF by training on a unified RCF data set with 341 data points covering 72 chemicals. We demonstrate the efficacy of the proposed gradient boosting regression tree (GBRT) model based on the extended connectivity fingerprints (ECFP) by predicting RCF values and achieved prediction performance with R-squared of 0.77 and mean absolute error (MAE) of 0.22 using 5-fold cross validation. In addition, our results reveal nonlinear relationships among properties of chemical, soil, and plant. Further in-depth analyses identify the key chemical topological substructures (e.g., -O, -Cl, aromatic rings and large conjugated π systems) related to RCF. Stemming from its simplicity and universality, the GBRT-ECFP model provides a valuable tool for LCIA and other environmental assessments to better characterize chemical risks to human health and ecosystems.

Development of an in silico prediction model for chemical-induced urinary tract toxicity by using naïve Bayes classifier.

The urinary tract toxicity is one of the major reasons for investigational drugs not coming into the market and even marketed drugs being restricted or withdrawn. The objective of this investigation is to develop an easily interpretable and practically applicable in silico prediction model of chemical-induced urinary tract toxicity by using naïve Bayes classifier. The genetic algorithm was used to select important molecular descriptors related to urinary tract toxicity, and the ECFP-6 fingerprint descriptors were applied to the urinary tract toxic/non-toxic fragments production. The established naïve Bayes classifier (NB-2) produced 87.3% overall accuracy of fivefold cross-validation for the training set and 84.2% for the external test set, which can be employed for the chemical-induced urinary tract toxicity assessment. Furthermore, six important molecular descriptors (e.g., number of N atoms, AlogP, molecular weight, number of H acceptors, number of H donors and molecular fractional polar surface area) and toxic and non-toxic fragments were obtained, which would help medicinal chemists interpret the mechanisms of urinary tract toxicity, and even provide theoretical guidance for hit and lead optimization.

Predicting drug-induced liver injury in human with Naïve Bayes classifier approach.

Drug-induced liver injury (DILI) is one of the major safety concerns in drug development. Although various toxicological studies assessing DILI risk have been developed, these methods were not sufficient in predicting DILI in humans. Thus, developing new tools and approaches to better predict DILI risk in humans has become an important and urgent task. In this study, we aimed to develop a computational model for assessment of the DILI risk with using a larger scale human dataset and Naïve Bayes classifier. The established Naïve Bayes prediction model was evaluated by 5-fold cross validation and an external test set. For the training set, the overall prediction accuracy of the 5-fold cross validation was 94.0 %. The sensitivity, specificity, positive predictive value and negative predictive value were 97.1, 89.2, 93.5 and 95.1 %, respectively. The test set with the concordance of 72.6 %, sensitivity of 72.5 %, specificity of 72.7 %, positive predictive value of 80.4 %, negative predictive value of 63.2 %. Furthermore, some important molecular descriptors related to DILI risk and some toxic/non-toxic fragments were identified. Thus, we hope the prediction model established here would be employed for the assessment of human DILI risk, and the obtained molecular descriptors and substructures should be taken into consideration in the design of new candidate compounds to help medicinal chemists rationally select the chemicals with the best prospects to be effective and safe.

In silico prediction of drug-induced myelotoxicity by using Naïve Bayes method.

Drug-induced myelotoxicity usually leads to decrease the production of platelets, red cells, and white cells. Thus, early identification and characterization of myelotoxicity hazard in drug development is very necessary. The purpose of this investigation was to develop a prediction model of drug-induced myelotoxicity by using a Naïve Bayes classifier. For comparison, other prediction models based on support vector machine and single-hidden-layer feed-forward neural network  methods were also established. Among all the prediction models, the Naïve Bayes classification model showed the best prediction performance, which offered an average overall prediction accuracy of [Formula: see text] for the training set and [Formula: see text] for the external test set. The significant contributions of this study are that we first developed a Naïve Bayes classification model of drug-induced myelotoxicity adverse effect using a larger scale dataset, which could be employed for the prediction of drug-induced myelotoxicity. In addition, several important molecular descriptors and substructures of myelotoxic compounds have been identified, which should be taken into consideration in the design of new candidate compounds to produce safer and more effective drugs, ultimately reducing the attrition rate in later stages of drug development.

Exploring QSAR models for activity-cliff prediction.

INTRODUCTION AND METHODOLOGY: Pairs of similar compounds that only differ by a small structural modification but exhibit a large difference in their binding affinity for a given target are known as activity cliffs (ACs). It has been hypothesised that QSAR models struggle to predict ACs and that ACs thus form a major source of prediction error. However, the AC-prediction power of modern QSAR methods and its quantitative relationship to general QSAR-prediction performance is still underexplored. We systematically construct nine distinct QSAR models by combining three molecular representation methods (extended-connectivity fingerprints, physicochemical-descriptor vectors and graph isomorphism networks) with three regression techniques (random forests, k-nearest neighbours and multilayer perceptrons); we then use each resulting model to classify pairs of similar compounds as ACs or non-ACs and to predict the activities of individual molecules in three case studies: dopamine receptor D2, factor Xa, and SARS-CoV-2 main protease. RESULTS AND CONCLUSIONS: Our results provide strong support for the hypothesis that indeed QSAR models frequently fail to predict ACs. We observe low AC-sensitivity amongst the evaluated models when the activities of both compounds are unknown, but a substantial increase in AC-sensitivity when the actual activity of one of the compounds is given. Graph isomorphism features are found to be competitive with or superior to classical molecular representations for AC-classification and can thus be employed as baseline AC-prediction models or simple compound-optimisation tools. For general QSAR-prediction, however, extended-connectivity fingerprints still consistently deliver the best performance amongs the tested input representations. A potential future pathway to improve QSAR-modelling performance might be the development of techniques to increase AC-sensitivity.

The applications of machine learning to predict the forming of chemically stable amorphous solid dispersions prepared by hot-melt extrusion.

Amorphous solid dispersion (ASD) is one of the most important strategies to improve the solubility and dissolution rate of poorly water-soluble drugs. As a widely used technique to prepare ASDs, hot-melt extrusion (HME) provides various benefits, including a solvent-free process, continuous manufacturing, and efficient mixing compared to solvent-based methods, such as spray drying. Energy input, consisting of thermal and specific mechanical energy, should be carefully controlled during the HME process to prevent chemical degradation and residual crystallinity. However, a conventional ASD development process uses a trial-and-error approach, which is laborious and time-consuming. In this study, we have successfully built multiple machine learning (ML) models to predict the amorphization of crystalline drug formulations and the chemical stability of subsequent ASDs prepared by the HME process. We utilized 760 formulations containing 49 active pharmaceutical ingredients (APIs) and multiple types of excipients. By evaluating the built ML models, we found that ECFP-LightGBM was the best model to predict amorphization with an accuracy of 92.8%. Furthermore, ECFP-XGBoost was the best in estimating chemical stability with an accuracy of 96.0%. In addition, the feature importance analyses based on SHapley Additive exPlanations (SHAP) and information gain (IG) revealed that several processing parameters and material attributes (i.e., drug loading, polymer ratio, drug's Extended-connectivity fingerprints (ECFP) fingerprints, and polymer's properties) are critical for achieving accurate predictions for the selected models. Moreover, important API's substructures related to amorphization and chemical stability were determined, and the results are largely consistent with the literature. In conclusion, we established the ML models to predict formation of chemically stable ASDs and identify the critical attributes during HME processing. Importantly, the developed ML methodology has the potential to facilitate the product development of ASDs manufactured by HME with a much reduced human workload.

Predicting pesticide dissipation half-life intervals in plants with machine learning models.

Pesticide dissipation half-life in plants is an important factor to assessing environmental fate of pesticides and establishing pre-harvest intervals critical to good agriculture practices. However, empirically measured pesticide dissipation half-lives are highly variable and the accurate prediction with models is challenging. This study utilized a dataset of pesticide dissipation half-lives containing 1363 datapoints, 311 pesticides, 10 plant types, and 4 plant component classes. Novel dissipation half-life intervals were proposed and predicted to account for high variations in empirical data. Four machine learning models (i.e., gradient boosting regression tree [GBRT], random forest [RF], supporting vector classifier [SVC], and logistic regression [LR]) were developed to predict dissipation half-life intervals using extended connectivity fingerprints (ECFP), temperature, plant type, and plant component class as model inputs. GBRT-ECFP had the best model performance with F1-microbinary score of 0.698 ± 0.010 for the binary classification compared with other machine learning models (e.g., LR-ECFP, F1-microbinary= 0.662 ± 0.009). Feature importance analysis of molecular structures in the binary classification identified aromatic rings, carbonyl group, organophosphate, =C-H, and N-containing heterocyclic groups as important substructures related to pesticide dissipation half-lives. This study suggests the utility of machine learning models in assessing the environmental fate of pesticides in agricultural crops.

A review on machine learning approaches and trends in drug discovery.

Drug discovery aims at finding new compounds with specific chemical properties for the treatment of diseases. In the last years, the approach used in this search presents an important component in computer science with the skyrocketing of machine learning techniques due to its democratization. With the objectives set by the Precision Medicine initiative and the new challenges generated, it is necessary to establish robust, standard and reproducible computational methodologies to achieve the objectives set. Currently, predictive models based on Machine Learning have gained great importance in the step prior to preclinical studies. This stage manages to drastically reduce costs and research times in the discovery of new drugs. This review article focuses on how these new methodologies are being used in recent years of research. Analyzing the state of the art in this field will give us an idea of where cheminformatics will be developed in the short term, the limitations it presents and the positive results it has achieved. This review will focus mainly on the methods used to model the molecular data, as well as the biological problems addressed and the Machine Learning algorithms used for drug discovery in recent years.

Mapping multiple principles of parietal-frontal cortical organization using functional connectivity.

Resting state functional connectivity has been promoted as a promising tool for creating cortical maps that show remarkable similarity to those established by invasive histological methods. While this tool has been largely used to identify and map cortical areas, its true potential in the context of studying connectional architecture and in conducting comparative neuroscience has remained unexplored. Here, we employ widely used resting state connectivity and data-driven clustering methods to extend this approach for the study of the organizational principles of the macaque parietal-frontal system. We show multiple, overlapping principles of organization, including a dissociation between dorsomedial and dorsolateral pathways and separate parietal-premotor and parietal-frontal pathways. These results demonstrate the suitability of this approach for understanding the complex organizational principles of the brain and for large-scale comparative neuroscience.

Identification of Essential 2D and 3D Chemical Features for Discovery of the Novel Tubulin Polymerization Inhibitors.

BACKGROUND: Tubulin polymerization inhibitors interfere with microtubule assembly and their functions lead to mitotic arrest, therefore they are attractive target for design and development of novel anticancer compounds. OBJECTIVE: The proposed novel and effective structures following the use of three-dimensionalquantitative structure activity relationship (3D-QSAR) pharmacophore based virtual screening clearly demonstrate the high efficiency of this method in modern drug discovery. METHODS: Combined computational approach was applied to extract the essential 2D and 3D features requirements for higher activity as well as identify new anti-tubulin agents. RESULTS: The best quantitative pharmacophore model, Hypo1, exhibited good correlation of 0.943 (RMSD=1.019) and excellent predictive power in the training set compounds. Generated model AHHHR, was well mapped to colchicine site and three-dimensional spatial arrangement of their features were in good agreement with the vital interactions in the active site. Total prediction accuracy (0.92 for training set and 0.86 for test set), enrichment factor (4.2 for training set and 4.5 for test set) and the area under the ROC curve (0.86 for training set and 0.94 for the test set), the developed model using Extended Class FingerPrints of maximum diameter 4 (ECFP_4) was chosen as the best model. CONCLUSION: Developed computational platform provided a better understanding of requirement features for colchicine site inhibitors and we believe the results of this study might be useful for the rational design and optimization of new inhibitors.

Development and evaluation of in silico prediction model for drug-induced respiratory toxicity by using naïve Bayes classifier method.

Respiratory toxicity is considered as main cause of drug withdrawal, which could seriously injure human health or even lead to death. The objective of this investigation was to develop an in silico prediction model of drug-induced respiratory toxicity by using naïve Bayes classifier. The genetic algorithm was used to select important molecular descriptors related to respiratory toxicity, and the ECFP_6 fingerprint descriptors were applied to the respiratory toxic/non-toxic fragments production. The established prediction model was validated by the internal 5-fold cross validation and external test set. The naïve Bayes classifier generated overall prediction accuracy of 91.8% for the training set and 84.3% for the external test set. Furthermore, six molecular descriptors (e.g., number of O atoms, number of N atoms, molecular weight, Apol, number of H acceptors and molecular polar surface area) considered as important for the drug-induced respiratory toxicity were identified, and some critical fragments related to the respiratory toxicity were achieved. We hope the established naïve Bayes prediction model could be used as a toxicological screening of chemicals for respiratory sensitization potential in drug development, and these obtained important information of respiratory toxic chemical structures could offer theoretical guidance for hit and lead optimization.

Development of novel in silico model for developmental toxicity assessment by using naïve Bayes classifier method.

Toxicological testing associated with developmental toxicity endpoints are very expensive, time consuming and labor intensive. Thus, developing alternative approaches for developmental toxicity testing is an important and urgent task in the drug development filed. In this investigation, the naïve Bayes classifier was applied to develop a novel prediction model for developmental toxicity. The established prediction model was evaluated by the internal 5-fold cross validation and external test set. The overall prediction results for the internal 5-fold cross validation of the training set and external test set were 96.6% and 82.8%, respectively. In addition, four simple descriptors and some representative substructures of developmental toxicants were identified. Thus, we hope the established in silico prediction model could be used as alternative method for toxicological assessment. And these obtained molecular information could afford a deeper understanding on the developmental toxicants, and provide guidance for medicinal chemists working in drug discovery and lead optimization.

Study of Structure-active Relationship for Inhibitors of HIV-1 Integrase LEDGF/p75 Interaction by Machine Learning Methods.

HIV-1 integrase (IN) is a promising target for anti-AIDS therapy, and LEDGF/p75 is proved to enhance the HIV-1 integrase strand transfer activity in vitro. Blocking the interaction between IN and LEDGF/p75 is an effective way to inhibit HIV replication infection. In this work, 274 LEDGF/p75-IN inhibitors were collected as the dataset. Support Vector Machine (SVM), Decision Tree (DT), Function Tree (FT) and Random Forest (RF) were applied to build several computational models for predicting whether a compound is an active or weakly active LEDGF/p75-IN inhibitor. Each compound is represented by MACCS fingerprints and CORINA Symphony descriptors. The prediction accuracies for the test sets of all the models are over 70 %. The best model Model 3B built by FT obtained a prediction accuracy and a Matthews Correlation Coefficient (MCC) of 81.08 % and 0.62 on test set, respectively. We found that the hydrogen bond and hydrophobic interactions are important for the bioactivity of an inhibitor.

Novel naïve Bayes classification models for predicting the chemical Ames mutagenicity.

Prediction of drug candidates for mutagenicity is a regulatory requirement since mutagenic compounds could pose a toxic risk to humans. The aim of this investigation was to develop a novel prediction model of mutagenicity by using a naïve Bayes classifier. The established model was validated by the internal 5-fold cross validation and external test sets. For comparison, the recursive partitioning classifier prediction model was also established and other various reported prediction models of mutagenicity were collected. Among these methods, the prediction performance of naïve Bayes classifier established here displayed very well and stable, which yielded average overall prediction accuracies for the internal 5-fold cross validation of the training set and external test set I set were 89.1±0.4% and 77.3±1.5%, respectively. The concordance of the external test set II with 446 marketed drugs was 90.9±0.3%. In addition, four simple molecular descriptors (e.g., Apol, No. of H donors, Num-Rings and Wiener) related to mutagenicity and five representative substructures of mutagens (e.g., aromatic nitro, hydroxyl amine, nitroso, aromatic amine and N-methyl-N-methylenemethanaminum) produced by ECFP_14 fingerprints were identified. We hope the established naïve Bayes prediction model can be applied to risk assessment processes; and the obtained important information of mutagenic chemicals can guide the design of chemical libraries for hit and lead optimization.

Development of novel prediction model for drug-induced mitochondrial toxicity by using naïve Bayes classifier method.

Mitochondrial dysfunction has been considered as an important contributing factor in the etiology of drug-induced organ toxicity, and even plays an important role in the pathogenesis of some diseases. The objective of this investigation was to develop a novel prediction model of drug-induced mitochondrial toxicity by using a naïve Bayes classifier. For comparison, the recursive partitioning classifier prediction model was also constructed. Among these methods, the prediction performance of naïve Bayes classifier established here showed best, which yielded average overall prediction accuracies for the internal 5-fold cross validation of the training set and external test set were 95 ± 0.6% and 81 ± 1.1%, respectively. In addition, four important molecular descriptors and some representative substructures of toxicants produced by ECFP_6 fingerprints were identified. We hope the established naïve Bayes prediction model can be employed for the mitochondrial toxicity assessment, and these obtained important information of mitochondrial toxicants can provide guidance for medicinal chemists working in drug discovery and lead optimization.

Novel naïve Bayes classification models for predicting the carcinogenicity of chemicals.

The carcinogenicity prediction has become a significant issue for the pharmaceutical industry. The purpose of this investigation was to develop a novel prediction model of carcinogenicity of chemicals by using a naïve Bayes classifier. The established model was validated by the internal 5-fold cross validation and external test set. The naïve Bayes classifier gave an average overall prediction accuracy of 90 ± 0.8% for the training set and 68 ± 1.9% for the external test set. Moreover, five simple molecular descriptors (e.g., AlogP, Molecular weight (MW), No. of H donors, Apol and Wiener) considered as important for the carcinogenicity of chemicals were identified, and some substructures related to the carcinogenicity were achieved. Thus, we hope the established naïve Bayes prediction model could be applied to filter early-stage molecules for this potential carcinogenicity adverse effect; and the identified five simple molecular descriptors and substructures of carcinogens would give a better understanding of the carcinogenicity of chemicals, and further provide guidance for medicinal chemists in the design of new candidate drugs and lead optimization, ultimately reducing the attrition rate in later stages of drug development.

Using Support Vector Machine (SVM) for Classification of Selectivity of H1N1 Neuraminidase Inhibitors.

Inhibition of the neuraminidase is one of the most promising strategies for preventing influenza virus spreading. 479 neuraminidase inhibitors are collected for dataset 1 and 208 neuraminidase inhibitors for A/P/8/34 are collected for dataset 2. Using support vector machine (SVM), four computational models were built to predict whether a compound is an active or weakly active inhibitor of neuraminidase. Each compound is represented by MASSC fingerprints and ADRIANA.Code descriptors. The predication accuracies for the test sets of all the models are over 78 %. Model 2B, which is the best model, obtains a prediction accuracy and a Matthews Correlation Coefficient (MCC) of 89.71 % and 0.81 on test set, respectively. The molecular polarizability, molecular shape, molecular size and hydrogen bonding are related to the activities of neuraminidase inhibitors. The models can be obtained from the authors.

In Silico Models to Discriminate Compounds Inducing and Noninducing Toxic Myopathy.

Toxic myopathy is a muscular disease in which the muscle fibers do not function and which results in muscular weakness. Some drugs, such as lipid-lowering drugs and antihistamines, can cause toxic myopathy. In this work, a dataset containing 232 chemical compounds inducing toxic myopathy (IM-compounds) and 117 drugs not inducing toxic myopathy (notIM-compounds) was collected. The dataset was split into a training set (containing 270 compounds) and a test set (containing 79 compounds). A Kohonen's self-organizing map (SOM) and a support vector machine (SVM) were applied to develop classification models to differentiate IM-compounds and notIM-compounds. Polarizibity related descriptors, electronegativity related descriptors, atom charges related descriptors, H-bonding related descriptor, atom identity and molecular shape descriptors were used to build models. Using the SOM method, classification accuracies of 88.4 % for the training set and 88.2 % for the test set were achieved; using the SVM method, classification accuracies of 95.6 % for the training set and 86.1 % for the test set were achieved. In addition, extended connectivity fingerprints (ECFP_4) were calculated and analyzed to find important substructures of molecules relating to toxic myopathy.