Comparing multilabel classification methods for provisional biopharmaceutics class prediction

Reliable Prediction of Caco-2 Permeability by Supervised Recursive Machine Learning Approaches

The heterogeneity of the Caco-2 cell line and differences in experimental protocols for permeability assessment using this cell-based method have resulted in the high variability of Caco-2 permeability measurements. These problems have limited the generation of large datasets to develop accurate and applicable regression models. This study presents a QSPR approach developed on the KNIME analytical platform and based on a structurally diverse dataset of over 4900 molecules. Interpretable models were obtained using random forest supervised recursive algorithms for data cleaning and feature selection. The development of a conditional consensus model based on regional and global regression random forest produced models with RMSE values between 0.43–0.51 for all validation sets. The potential applicability of the model as a surrogate for the in vitro Caco-2 assay was demonstrated through blind prediction of 32 drugs recommended by the International Council for the Harmonization of Technical Requirements for Pharmaceuticals (ICH) for validation of in vitro permeability methods. The model was validated for the preliminary estimation of the BCS/BDDCS class. The KNIME workflow developed to automate new drug prediction is freely available. The results suggest that this automated prediction platform is a reliable tool for identifying the most promising compounds with high intestinal permeability during the early stages of drug discovery.

Multi-label classification and label dependence in in silico toxicity prediction

Most computational predictive models are specifically trained for a single toxicity endpoint and lack the ability to learn dependencies between endpoints, such as those targeting similar biological pathways. In this study, we compare the performance of 3 multi-label classification (MLC) models, namely Classifier Chains (CC), Label Powersets (LP) and Stacking (SBR), against independent classifiers (Binary Relevance) on Tox21 challenge data. Also, we develop a novel label dependence measure that shows full range of values, even at low prior probabilities, for the purpose of data-driven label partitioning. Using Logistic Regression as the base classifier and random label partitioning (k = 3), CC show statistically significant improvements in model performance using Hamming and multi-label accuracy scores (p<0.05), while SBR show significant improvements in multi-label accuracy scores. The weights in the Logistic Regression and Stacking models are positively associated with label dependencies, suggesting that learning label dependence is a key contributor to improving model performance. An original quantitative measure of label dependency is combined with the Louvain community detection method to learn label partitioning using a data-driven process. The resulting MLCs with learned label partitioning were generally found to be non-inferior to their corresponding random or no label partitioning counterparts. Additionally, using the Random Forest classifier in a 10-fold stratified cross validation Stacking model, we find that the top-performing stacking model out-performs the corresponding base model in 11 out of 12 Tox21 labels. Taken together, these results suggest that MLC models could potentially boost the performance of current single-endpoint predictive models and that label partitioning learning may be used in place of random label partitionings.

Toward a hemorrhagic trauma severity score: fusing five physiological biomarkers

BACKGROUND

To introduce the Hemorrhage Intensive Severity and Survivability (HISS) score, based on the fusion of multi-biomarker data; glucose, lactate, pH, potassium, and oxygen tension, to serve as a patient-specific attribute in hemorrhagic trauma.

MATERIALS AND METHODS

One hundred instances of Sensible Fictitious Rationalized Patient (SFRP) data were synthetically generated and the HISS score assigned by five clinically active physician experts (100 [5]). The HISS score stratifies the criticality of the trauma patient as; low(0), guarded(1), elevated(2), high(3) and severe(4). Standard classifier algorithms; linear support vector machine (SVM-L), multi-class ensemble bagged decision tree (EBDT), artificial neural network with bayesian regularization (ANN:BR) and possibility rule-based using function approximation (PRBF) were evaluated for their potential to similarly classify and predict a HISS score.

RESULTS

SVM-L, EBDT, ANN:BR and PRBF generated score predictions with testing accuracies (majority vote) corresponding to 0.91 ± 0.06, 0.93 ± 0.04, 0.92 ± 0.07, and 0.92 ± 0.03, respectively, with no statistically significant difference (p > 0.05). Targeted accuracies of 0.99 and 0.999 could be achieved with SFRP data size and clinical expert scores of 147[7](0.99) and 154[9](0.999), respectively.

CONCLUSIONS

The predictions of the data-driven model in conjunction with an adjunct multi-analyte biosensor intended for point-of-care continual monitoring of trauma patients, can aid in patient stratification and triage decision-making.

An Investigation into the Factors Governing Drug Absorption and Food Effect Prediction Based on Data Mining Methodology

Drug absorption is a complex process governed by a number of interrelated physicochemical, biopharmaceutical, and pharmacokinetic factors. In order to explore complex relationships among these factors, multivariate exploratory analysis was performed on the dataset of drugs with diverse bioperformance. The investigated dataset included subset of drugs for which bioequivalence between solid dosage form and oral solution has been reported, and subset of drugs described in the literature as low solubility/low permeability compounds. Discriminatory power of hierarchical clustering on principal components was somewhat higher when applied on the data subsets of drugs with similar bioperformance, while analysis of the integrated dataset indicated existence of two groups of drugs with the boundaries reflected in P_eff value of approximately 2 × 10^-4 cm/s and Fa and Fm values higher than 85% and 50%, respectively. Majority of the investigated drugs within the integrated dataset were grouped within their initial subset indicating that overall drug bioperformance is closely related to its physicochemical, biopharmaceutical and pharmacokinetic properties. Classification models constructed using the random forest (RF) and support vector machine with polynomial kernel function were able to predict food effect based on drug dose/solubility ratio (D/S), effective permeability (P_eff), percent of dose metabolized (Fm), and elimination half-life (τ_1/2). Although both models performed well during training and testing, only RF kept satisfying performance when applied on the external dataset (kappa value > 0.4). The results obtained indicate that data mining can be employed as useful tool in biopharmaceutical drug characterization which merits further investigation.

Multilabel and Missing Label Methods for Binary Quantitative Structure-Activity Relationship Models: An Application for the Prediction of Adverse Drug Reactions

The prediction of adverse drug reactions in the discovery of new medicines is highly challenging. In the task of predicting the adverse reactions of chemical compounds, information about different targets is often available. Although we can focus on every adverse drug reaction prediction separately, multilabel approaches have been proven useful in many research areas for taking advantage of the relationship among the targets. However, when approaching the prediction problem from a multilabel point of view, we have to deal with the lack of information for some labels. This missing labels problem is a relevant issue in the field of cheminformatics approaches. This paper aims to predict the adverse drug reaction of commercial drugs using a multilabel approach where the possible presence of missing labels is also taken into consideration. We propose the use of multilabel methods to deal with the prediction of a large set of 27 different adverse reaction targets. We also propose the use of multilabel methods specifically designed to deal with the missing labels problem to test their ability to solve this difficult problem. The results show the validity of the proposed approach, demonstrating a superior performance of the multilabel method compared with the single-label approach in addressing the problem of adverse drug reaction prediction.

In Silico Assessment of ADME Properties: Advances in Caco-2 Cell Monolayer Permeability Modeling

One of the main goals of in silico Caco-2 cell permeability models is to identify those drug substances with high intestinal absorption in human (HIA). For more than a decade, several in silico Caco-2 models have been made, applying a wide range of modeling techniques; nevertheless, their capacity for intestinal absorption extrapolation is still doubtful. There are three main problems related to the modest capacity of obtained models, including the existence of inter- and/or intra-laboratory variability of recollected data, the influence of the metabolism mechanism, and the inconsistent in vitro-in vivo correlation (IVIVC) of Caco-2 cell permeability. This review paper intends to sum up the recent advances and limitations of current modeling approaches, and revealed some possible solutions to improve the applicability of in silico Caco-2 permeability models for absorption property profiling, taking into account the above-mentioned issues.

Application of data mining approach to identify drug subclasses based on solubility and permeability

Solubility and permeability are recognized as key parameters governing drug intestinal absorption and represent the basis for biopharmaceutics drug classification. The Biopharmaceutics Classification System (BCS) is widely accepted and adopted by regulatory agencies. However, currently established low/high permeability and solubility boundaries are the subject of the ongoing scientific discussion. The aim of the present study was to apply data mining analysis on the selected drugs data set in order to develop a human permeability predictive model based on selected molecular descriptors, and to perform data clustering and classification to identify drug subclasses with respect to dose/solubility ratio (D/S) and effective permeability (P_eff ). The P_eff values predicted for 30 model drugs for which experimental human permeability data are not available were in good agreement with the reported fraction of drug absorbed. The results of clustering and classification analysis indicate the predominant influence of P_eff over D/S. Two P_eff cut-off values (1 × 10^-4 and 2.7 × 10^-4  cm/s) have been identified indicating the existence of an intermediate group of drugs with moderate permeability. Advanced computational analysis employed in the present study enabled the recognition of complex relationships and patterns within physicochemical and biopharmaceutical properties associated with drug bioperformance.

Integrating theoretical and experimental permeability estimations for provisional biopharmaceutical classification: Application to the WHO essential medicines

The accuracy of the provisional estimation of the Biopharmaceutics Classification System (BCS) is heavily influenced by the permeability measurement. In this study, several theoretical and experimental models currently employed for BCS permeability classification have been analysed. The experimental models included the in situ rat intestinal perfusion, the ex vivo rat intestinal tissue in an Ussing chamber, the MDCK and Caco-2 cell monolayers, and the parallel artificial membrane (PAMPA). The theoretical models included the octanol-water partition coefficient and the QSPeR (Quantitative Structure-Permeability Relationship) model recently developed. For model validation, a dataset of 43 compounds has been recompiled and analysed for the suitability for BCS permeability classification in comparison with the use of human intestinal absorption and oral bioavailability values. The application of the final model, based on a majority voting system showed a 95.3% accuracy for predicting human permeability. Finally, the present approach was applied to the 186 orally administered drugs in immediate-release dosage forms of the WHO Model List of Essential Medicines. The percentages of the drugs that were provisionally classified as BCS Class I and Class III was 62.4%, suggesting that in vivo bioequivalence (BE) may potentially be assured with a less expensive and more easily implemented in vitro dissolution test, ensuring the efficiency and quality of pharmaceutical products. The results of the current study improve the accuracy of provisional BCS classification by combining different permeability models.

Deep Learning Applications for Predicting Pharmacological Properties of Drugs and Drug Repurposing Using Transcriptomic Data

Deep learning is rapidly advancing many areas of science and technology with multiple success stories in image, text, voice and video recognition, robotics, and autonomous driving. In this paper we demonstrate how deep neural networks (DNN) trained on large transcriptional response data sets can classify various drugs to therapeutic categories solely based on their transcriptional profiles. We used the perturbation samples of 678 drugs across A549, MCF-7, and PC-3 cell lines from the LINCS Project and linked those to 12 therapeutic use categories derived from MeSH. To train the DNN, we utilized both gene level transcriptomic data and transcriptomic data processed using a pathway activation scoring algorithm, for a pooled data set of samples perturbed with different concentrations of the drug for 6 and 24 hours. In both pathway and gene level classification, DNN achieved high classification accuracy and convincingly outperformed the support vector machine (SVM) model on every multiclass classification problem, however, models based on pathway level data performed significantly better. For the first time we demonstrate a deep learning neural net trained on transcriptomic data to recognize pharmacological properties of multiple drugs across different biological systems and conditions. We also propose using deep neural net confusion matrices for drug repositioning. This work is a proof of principle for applying deep learning to drug discovery and development.