Predictive Capability of QSAR Models Based on the CompTox Zebrafish Embryo Assays: An Imbalanced Classification Problem.
Ontology highlight
ABSTRACT:
SUBMITTER: Lovric M
PROVIDER: S-EPMC7998177 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Drug-induced liver injury is a major concern in the drug development process. Expensive and time-consuming in vitro and in vivo studies do not reflect the complexity of the phenomenon. Complementary to wet lab methods are in silico approaches, which present a cost-efficient method for toxicity prediction. The aim of our study was to explore the capabilities of counter-propagation artificial neural networks (CPANNs) for the classification of an imbalanced dataset related to idiosyncratic drug-induced liver injury and to develop a model for prediction of the hepatotoxic potential of drugs. Genetic algorithm optimization of CPANN models was used to build models for the classification of drugs into hepatotoxic and non-hepatotoxic class using molecular descriptors. For the classification of an imbalanced dataset, we modified the classical CPANN training algorithm by integrating random subsampling into the training procedure of CPANN to improve the classification ability of CPANN. According to the number of models accepted by internal validation and according to the prediction statistics on the external set, we concluded that using an imbalanced set with balanced subsampling in each learning epoch is a better approach compared to using a fixed balanced set in the case of the counter-propagation artificial neural network learning methodology.
| S-EPMC7037161 | biostudies-literature
Project description:Several hundred disinfection byproducts (DBPs) in drinking water have been identified, and are known to have potentially adverse health effects. There are toxicological data gaps for most DBPs, and the predictive method may provide an effective way to address this. The development of an in-silico model of toxicology endpoints of DBPs is rarely studied. The main aim of the present study is to develop predictive quantitative structure-activity relationship (QSAR) models for the reactive toxicities of 50 DBPs in the five bioassays of X-Microtox, GSH+, GSH-, DNA+ and DNA-. All-subset regression was used to select the optimal descriptors, and multiple linear-regression models were built. The developed QSAR models for five endpoints satisfied the internal and external validation criteria: coefficient of determination (R²) > 0.7, explained variance in leave-one-out prediction (Q²LOO) and in leave-many-out prediction (Q²LMO) > 0.6, variance explained in external prediction (Q²F1, Q²F2, and Q²F3) > 0.7, and concordance correlation coefficient (CCC) > 0.85. The application domains and the meaning of the selective descriptors for the QSAR models were discussed. The obtained QSAR models can be used in predicting the toxicities of the 50 DBPs.
| S-EPMC6151816 | biostudies-literature
Project description:Validation is a crucial aspect of quantitative structure-activity relationship (QSAR) modeling. The present paper shows that traditionally used validation parameters (leave-one-out Q(2) for internal validation and predictive R(2) for external validation) may be supplemented with two novel parameters r(m)(2) and R(p)(2) for a stricter test of validation. The parameter r(m)(2)((overall)) penalizes a model for large differences between observed and predicted values of the compounds of the whole set (considering both training and test sets) while the parameter R(p)(2) penalizes model R(2) for large differences between determination coefficient of nonrandom model and square of mean correlation coefficient of random models in case of a randomization test. Two other variants of r(m)(2) parameter, r(m)(2)((LOO)) and r(m)(2)((test)), penalize a model more strictly than Q(2) and R(2)(pred) respectively. Three different data sets of moderate to large size have been used to develop multiple models in order to indicate the suitability of the novel parameters in QSAR studies. The results show that in many cases the developed models could satisfy the requirements of conventional parameters (Q(2) and R(2)(pred)) but fail to achieve the required values for the novel parameters r(m)(2) and R(p)(2). Moreover, these parameters also help in identifying the best models from among a set of comparable models. Thus, a test for these two parameters is suggested to be a more stringent requirement than the traditional validation parameters to decide acceptability of a predictive QSAR model, especially when a regulatory decision is involved.
| S-EPMC6254296 | biostudies-literature
Project description:Quantitative structure-activity relationships (QSAR) are a widely used methodology allowing not only a better understanding of the mechanisms of chemical reactions, including radical scavenging, but also to predict the relevant properties of chemical compounds without their synthesis, isolation and experimental testing. Unlike the QSAR modeling of the kinetic antioxidant assays, modeling of the assays with stoichiometric endpoints depends strongly on the number of hydroxyl groups in the antioxidant molecule, as well as on some integral molecular descriptors characterizing the proportion of OH-groups able to enter and complete the radical scavenging reaction. In this work, we tested the feasibility of a "hybrid" classification/regression approach, consisting of explicit classification of individual OH-groups as involved in radical scavenging reactions, and using further the number of these OH-groups as a descriptor in simple-regression QSAR models of antiradical capacity assays with stoichiometric endpoints. A simple threshold classification based on the sum of trolox-equivalent antiradical capacity values was used, selecting OH-groups with specific radical stability- and reactivity-related electronic parameters or their combination as "active" or "inactive". We showed that this classification/regression modeling approach provides a substantial improvement of the simple-regression QSAR models over those built on the number of total phenolic OH-groups only, and yields a statistical performance similar to that of the best reported multiple-regression QSARs for antiradical capacity assays with stoichiometric endpoints.
| S-EPMC9000788 | biostudies-literature
Project description:A key challenge in the field of Quantitative Structure Activity Relationships (QSAR) is how to effectively treat experimental error in the training and evaluation of computational models. It is often assumed in the field of QSAR that models cannot produce predictions which are more accurate than their training data. Additionally, it is implicitly assumed, by necessity, that data points in test sets or validation sets do not contain error, and that each data point is a population mean. This work proposes the hypothesis that QSAR models can make predictions which are more accurate than their training data and that the error-free test set assumption leads to a significant misevaluation of model performance. This work used 8 datasets with six different common QSAR endpoints, because different endpoints should have different amounts of experimental error associated with varying complexity of the measurements. Up to 15 levels of simulated Gaussian distributed random error was added to the datasets, and models were built on the error laden datasets using five different algorithms. The models were trained on the error laden data, evaluated on error-laden test sets, and evaluated on error-free test sets. The results show that for each level of added error, the RMSE for evaluation on the error free test sets was always better. The results support the hypothesis that, at least under the conditions of Gaussian distributed random error, QSAR models can make predictions which are more accurate than their training data, and that the evaluation of models on error laden test and validation sets may give a flawed measure of model performance. These results have implications for how QSAR models are evaluated, especially for disciplines where experimental error is very large, such as in computational toxicology.
| S-EPMC8613965 | biostudies-literature
Project description:3D printing has emerged as a rapid and cost-efficient manufacturing technique to enable the fabrication of bespoke, complex prototypes. If the technology is to have a significant impact in biomedical applications, such as drug discovery and molecular diagnostics, the devices produced must be biologically compatible to enable their use with established reference assays and protocols. In this work we demonstrate that we can adapt the Fish Embryo Test (FET) as a new method to quantify the toxicity of 3D printed microfluidic devices. We assessed the biocompatibility of four commercially available 3D printing polymers (VisiJetCrystal EX200, Watershed 11122XC, Fototec SLA 7150 Clear and ABSplus P-430), through the observation of key developmental markers in the developing zebrafish embryos. Results show all of the photopolymers to be highly toxic to the embryos, resulting in fatality, although we do demonstrate that post-printing treatment of Fototec 7150 makes it suitable for zebrafish culture within the FET.
| S-EPMC4758231 | biostudies-literature
Project description:Many of the structures in PubChem are annotated with activities determined in high-throughput screening (HTS) assays. Because of the nature of these assays, the activity data are typically strongly imbalanced, with a small number of active compounds contrasting with a very large number of inactive compounds. We have used several such imbalanced PubChem HTS assays to test and develop strategies to efficiently build robust QSAR models from imbalanced data sets. Different descriptor types [Quantitative Neighborhoods of Atoms (QNA) and "biological" descriptors] were used to generate a variety of QSAR models in the program GUSAR. The models obtained were compared using external test and validation sets. We also report on our efforts to incorporate the most predictive of our models in the publicly available NCI/CADD Group Web services ( http://cactus.nci.nih.gov/chemical/apps/cap).
| S-EPMC3985743 | biostudies-other
Project description:BackgroundBayesian models have been applied throughout the Covid-19 pandemic especially to model time series of case counts or deaths. Fewer examples exist of spatio-temporal modeling, even though the spatial spread of disease is a crucial factor in public health monitoring. The predictive capabilities of infectious disease models is also important.MethodsIn this study, the ability of Bayesian hierarchical models to recover different parts of the variation in disease counts is the focus. It is clear that different measures provide different views of behavior when models are fitted prospectively. Over a series of time horizons one step predictions have been generated and compared for different models (for case counts and death counts). These Bayesian SIR models were fitted using MCMC at 28 time horizons to mimic prospective prediction. A range of goodness of prediction measures were analyzed across the different time horizons.ResultsA particularly important result is that the peak intensity of case load is often under-estimated, while random spikes in case load can be mimicked using time dependent random effects. It is also clear that during the early wave of the pandemic simpler model forms are favored, but subsequently lagged spatial dependence models for cases are favored, even if the sophisticated models perform better overall.DiscussionThe models fitted mimic the situation where at a given time the history of the process is known but the future must be predicted based on the current evolution which has been observed. Using an overall 'best' model for prediction based on retrospective fitting of the complete pandemic waves is an assumption. However it is also clear that this case count model is well favored over other forms. During the first wave a simpler time series model predicts case counts better for counties than a spatially dependent one. The picture is more varied for morality.ConclusionsFrom a predictive point of view it is clear that spatio-temporal models applied to county level Covid-19 data within the US vary in how well they fit over time and also how well they predict future events. At different times, SIR case count models and also mortality models with cumulative counts perform better in terms of prediction. A fundamental result is that predictive capability of models varies over time and using the same model could lead to poor predictive performance. In addition it is clear that models addressing the spatial context for case counts (i.e. with lagged neighborhood terms) and cumulative case counts for mortality data are clearly better at modeling spatio-temporal data which is commonly available for the Covid-19 pandemic in different areas of the globe.
| S-EPMC10422743 | biostudies-literature
Project description:BackgroundNew methods of continuous glucose monitoring (CGM) provide real-time alerts for hypoglycemia, hyperglycemia, and rapid fluctuations of glucose levels, thereby improving glycemic control, which is especially crucial during meals and physical activity. However, complex CGM systems pose challenges for individuals with diabetes and healthcare professionals, particularly when interpreting rapid glucose level changes, dealing with sensor delays (approximately a 10 min difference between interstitial and plasma glucose readings), and addressing potential malfunctions. The development of advanced predictive glucose level classification models becomes imperative for optimizing insulin dosing and managing daily activities.MethodsThe aim of this study was to investigate the efficacy of three different predictive models for the glucose level classification: (1) an autoregressive integrated moving average model (ARIMA), (2) logistic regression, and (3) long short-term memory networks (LSTM). The performance of these models was evaluated in predicting hypoglycemia (<70 mg/dL), euglycemia (70-180 mg/dL), and hyperglycemia (>180 mg/dL) classes 15 min and 1 h ahead. More specifically, the confusion matrices were obtained and metrics such as precision, recall, and accuracy were computed for each model at each predictive horizon.ResultsAs expected, ARIMA underperformed the other models in predicting hyper- and hypoglycemia classes for both the 15 min and 1 h horizons. For the 15 min forecast horizon, the performance of logistic regression was the highest of all the models for all glycemia classes, with recall rates of 96% for hyper, 91% for norm, and 98% for hypoglycemia. For the 1 h forecast horizon, the LSTM model turned out to be the best for hyper- and hypoglycemia classes, achieving recall values of 85% and 87% respectively.ConclusionsOur findings suggest that different models may have varying strengths and weaknesses in predicting glucose level classes, and the choice of model should be carefully considered based on the specific requirements and context of the clinical application. The logistic regression model proved to be more accurate for the next 15 min, particularly in predicting hypoglycemia. However, the LSTM model outperformed logistic regression in predicting glucose level class for the next hour. Future research could explore hybrid models or ensemble approaches that combine the strengths of multiple models to further enhance the accuracy and reliability of glucose predictions.
| S-EPMC10574913 | biostudies-literature
Project description:The knowledge-based Toxtree expert system (SAR approach) was integrated with the statistically based counter propagation artificial neural network (CP ANN) model (QSAR approach) to contribute to a better mechanistic understanding of a carcinogenicity model for non-congeneric chemicals using Dragon descriptors and carcinogenic potency for rats as a response. The transparency of the CP ANN algorithm was demonstrated using intrinsic mapping technique specifically Kohonen maps. Chemical structures were represented by Dragon descriptors that express the structural and electronic features of molecules such as their shape and electronic surrounding related to reactivity of molecules. It was illustrated how the descriptors are correlated with particular structural alerts (SAs) for carcinogenicity with recognized mechanistic link to carcinogenic activity. Moreover, the Kohonen mapping technique enables one to examine the separation of carcinogens and non-carcinogens (for rats) within a family of chemicals with a particular SA for carcinogenicity. The mechanistic interpretation of models is important for the evaluation of safety of chemicals.
| S-EPMC3962111 | biostudies-literature