DemQSAR: predicting human volume of distribution and clearance of drugs

Predicting Elimination of Small-Molecule Drug Half-Life in Pharmacokinetics Using Ensemble and Consensus Machine Learning Methods

Half-life is a significant pharmacokinetic parameter included in the excretion phase of absorption, distribution, metabolism, and excretion. It is one of the key factors for the successful marketing of drug candidates. Therefore, predicting half-life is of great significance in drug design. In this study, we employed eXtreme Gradient Boosting (XGboost), randomForest (RF), gradient boosting machine (GBM), and supporting vector machine (SVM) to build quantitative structure-activity relationship (QSAR) models on 3512 compounds and evaluated model performance by using root-mean-square error (RMSE), R2, and mean absolute error (MAE) metrics and interpreted features by SHapley Additive exPlanation (SHAP). Furthermore, we developed consensus models through integrating four individual models and validated their performance using a Y-randomization test and applicability domain analysis. Finally, matched molecular pair analysis was used to extract the transformation rules. Our results revealed that XGboost outperformed other individual models (RMSE = 0.176, R2 = 0.845, MAE = 0.141). The consensus model integrating all four models continued to enhance prediction performance (RMSE = 0.172, R2 = 0.856, MAE = 0.138). We evaluated the reliability, robustness, and generalization ability via Y-randomization test and applicability domain analysis. Meanwhile, we utilized SHAP to interpret features and employed matched molecular pair analysis to extract chemical transformation rules that provide suggestions for optimizing drug structure. In conclusion, we believe that the consensus model developed in this study serve as a reliable tool to evaluate half-life in drug discovery, and the chemical transformation rules concluded in this study could provide valuable suggestions in drug discovery.

Prediction of Human Clearance Using In Silico Models with Reduced Bias

Predicting human clearance with high accuracy from in silico-derived parameters alone is highly desirable, as it is fast, saves in vitro resources, and is animal-sparing. We derived random forest (RF) models from 1340 compounds with human intravenous pharmacokinetic (PK) data, the largest data set publicly available today. To assess the general applicability of the RF models, we systematically removed structural-therapeutic class analogues and other compounds with structural similarity from the training sets. For a quasi-prospective test set of 343 compounds, we show that RF models devoid of structurally similar compounds in the training set predict human clearance with a geometric mean fold error (GMFE) of 3.3. While the observed GMFE illustrates how difficult it is to generate a useful model that is broadly applicable, we posit that our RF models yield a more realistic assessment of how well human clearance can be predicted prospectively. We deployed the conformal prediction formalism to assess the model applicability and to determine the prediction confidence intervals for each prediction. We observed that clearance can be predicted better for renally cleared compounds than for other clearance mechanisms. We show that applying a classification model for predicting renal clearance identifies a subset of compounds for which clearance can be predicted with higher accuracy, yielding a GMFE of 2.3. In addition, our in silico RF human clearance models compared well to models derived from scaling human hepatocytes or preclinical in vivo data.

Improving ADMET Prediction Accuracy for Candidate Drugs: Factors to Consider in QSPR Modeling Approaches

Quantitative Structure-Property Relationship (QSPR) employs mathematical and statistical methods to reveal quantitative correlations between the pharmacokinetics of compounds and their molecular structures, as well as their physical and chemical properties. QSPR models have been widely applied in the prediction of drug absorption, distribution, metabolism, excretion, and toxicity (ADMET). However, the accuracy of QSPR models for predicting drug ADMET properties still needs improvement. Therefore, this paper comprehensively reviews the tools employed in various stages of QSPR predictions for drug ADMET. It summarizes commonly used approaches to building QSPR models, systematically analyzing the advantages and limitations of each modeling method to ensure their judicious application. We provide an overview of recent advancements in the application of QSPR models for predicting drug ADMET properties. Furthermore, this review explores the inherent challenges in QSPR modeling while also proposing a range of considerations aimed at enhancing model prediction accuracy. The objective is to enhance the predictive capabilities of QSPR models in the field of drug development and provide valuable reference and guidance for researchers in this domain.

Application of machine learning techniques to the analysis and prediction of drug pharmacokinetics

In this review, we describe the current status and challenges in applying machine-learning techniques to the analysis and prediction of pharmacokinetic data. The theory of pharmacokinetics has been developed over decades on the basis of physiology and reaction kinetics. Mathematical models allow the reduction of pharmacokinetic data to parameter values, giving insight and understanding into ADME processes and predicting the outcome of different dosing scenarios. However, much information hidden in the data is lost through conceptual simplification with models. It is difficult to use mechanistic models alone to predict diverse pharmacokinetic time profiles, including inter-drug and inter-individual differences, in a cross-sectional manner. Machine learning is a prediction platform that can handle complex phenomena through data-driven analysis. As a resule, machine learning has been successfully adopted in various fields, including image recognition and language processing, and has been used for over two decades in pharmacokinetic research, primarily in the area of quantitative structure-activity relationships for pharmacokinetic parameters. Machine-learning models are generally known to provide better predictive performance than conventional linear models. Owing to the recent success in deep learning, models with new structures are being consistently proposed. These models include transfer learning and generative adversarial networks, which contribute to the effective use of a limited amount of data by diverting existing similar models or generating pseudo-data. How to make such newly emerging machine learning technologies applicable to meet challenges in the pharmacokinetics/pharmacodynamics field is now the key issue.

Predicting Total Drug Clearance and Volumes of Distribution Using the Machine Learning-Mediated Multimodal Method through the Imputation of Various Nonclinical Data

Pharmacokinetic research plays an important role in the development of new drugs. Accurate predictions of human pharmacokinetic parameters are essential for the success of clinical trials. Clearance (CL) and volume of distribution (Vd) are important factors for evaluating pharmacokinetic properties, and many previous studies have attempted to use computational methods to extrapolate these values from nonclinical laboratory animal models to human subjects. However, it is difficult to obtain sufficient, comprehensive experimental data from these animal models, and many studies are missing critical values. This means that studies using nonclinical data as explanatory variables can only apply a small number of compounds to their model training. In this study, we perform missing-value imputation and feature selection on nonclinical data to increase the number of training compounds and nonclinical datasets available for these kinds of studies. We could obtain novel models for total body clearance (CL_tot) and steady-state Vd (Vd_ss) (CL_tot: geometric mean fold error [GMFE], 1.92; percentage within 2-fold error, 66.5%; Vd_ss: GMFE, 1.64; percentage within 2-fold error, 71.1%). These accuracies were comparable to the conventional animal scale-up models. Then, this method differs from animal scale-up methods because it does not require animal experiments, which continue to become more strictly regulated as time passes.

Machine learning in pharmacometrics: Opportunities and challenges

The explosive growth in medical devices, imaging and diagnostics, computing, and communication and information technologies in drug development and healthcare has created an ever-expanding data landscape that the pharmacometrics (PMX) research community must now traverse. The tools of machine learning (ML) have emerged as a powerful computational approach in other data-rich disciplines but its effective utilization in the pharmaceutical sciences and PMX modelling is in its infancy. ML-based methods can complement PMX modelling by enabling the information in diverse sources of big data, e.g. population-based public databases and disease-specific clinical registries, to be harnessed because they are capable of efficiently identifying salient variables associated with outcomes and delineating their interdependencies. ML algorithms are computationally efficient, have strong predictive capabilities and can enable learning in the big data setting. ML algorithms can be viewed as providing a computational bridge from big data to complement PMX modelling. This review provides an overview of the strengths and weaknesses of ML approaches vis-à-vis population methods, assesses current research into ML applications in the pharmaceutical sciences and provides perspective for potential opportunities and strategies for the successful integration and utilization of ML in PMX.

In Silico Models of Human PK Parameters. Prediction of Volume of Distribution Using an Extensive Data Set and a Reduced Number of Parameters

A novel, descriptor-parsimonious in silico model to predict human VDss (volume of distribution at steady-state) has been derived and thoroughly tested in a quasi-prospective regimen using an independent test set of 213 compounds. The model performs on par with a former benchmark model that relied on far more descriptors. As a result, the new random forest model relying on only six descriptors allows for interpretations that help chemists to design compounds with desired human VDss values. A comparison of in silico predictions of VDss with models using in vitro derived descriptors or in vivo scaling methods supports the strength of the in-silico approach, considering its resource- and animal-sparing nature. The strong performance of the in silico VDss models on structurally novel compounds supports the high degree of confidence that can be placed in using in silico human VDss predictions for compound design and human dose predictions.

Therapeutic value of steroidal alkaloids in cancer: Current trends and future perspectives

Discovery and development of new potentially selective anticancer agents are necessary to prevent a global cancer health crisis. Currently, alternative medicinal agents derived from plants have been extensively investigated to develop anticancer drugs with fewer adverse effects. Among them, steroidal alkaloids are conventional secondary metabolites that comprise an important class of natural products found in plants, marine organisms and invertebrates, and constitute a judicious choice as potential anti-cancer leads. Traditional medicine and modern science have shown that representatives from this compound group possess potential antimicrobial, analgesic, anticancer and anti-inflammatory effects. Therefore, systematic and recapitulated information about the bioactivity of these compounds, with special emphasis on the molecular or cellular mechanisms, is of high interest. In this review, we methodically discuss the in vitro and in vivo potential of the anticancer activity of natural steroidal alkaloids and their synthetic and semi-synthetic derivatives. This review focuses on cumulative and comprehensive molecular mechanisms, which will help researchers understand the molecular pathways involving steroid alkaloids to generate a selective and safe new lead compound with improved therapeutic applications for cancer prevention and therapy. In vitro and in vivo studies provide evidence about the promising therapeutic potential of steroidal alkaloids in various cancer cell lines, but advanced pharmacokinetic and clinical experiments are required to develop more selective and safe drugs for cancer treatment.

Molecular Modeling Approaches for the Prediction of Selected Pharmacokinetic Properties

Poor profiles of potential drug candidates, including pharmacokinetic properties, have been acknowledged as a significant hindrance to the development of modern therapeutics. Contemporary drug discovery and development would be incomplete without the aid of molecular modeling (in-silico) techniques, allowing the prediction of pharmacokinetic properties such as clearance, unbound fraction, volume of distribution and bioavailability. As with all models, in-silico approaches are subject to their interpretability, a trait that must be balanced with accuracy when considering the development of new methods. The best models will always require reliable data to inform them, presenting significant challenges, particularly when appropriate in-vitro or in-vivo data may be difficult or time-consuming to obtain. This article seeks to review some of the key in-silico techniques used to predict key pharmacokinetic properties and give commentary on the current and future directions of the field.

An Integrated Transfer Learning and Multitask Learning Approach for Pharmacokinetic Parameter Prediction

BACKGROUND

Pharmacokinetic evaluation is one of the key processes in drug discovery and development. However, current absorption, distribution, metabolism, and excretion prediction models still have limited accuracy.

AIM

This study aims to construct an integrated transfer learning and multitask learning approach for developing quantitative structure-activity relationship models to predict four human pharmacokinetic parameters.

METHODS

A pharmacokinetic data set included 1104 U.S. FDA approved small molecule drugs. The data set included four human pharmacokinetic parameter subsets (oral bioavailability, plasma protein binding rate, apparent volume of distribution at steady-state, and elimination half-life). The pretrained model was trained on over 30 million bioactivity data entries. An integrated transfer learning and multitask learning approach was established to enhance the model generalization.

RESULTS

The pharmacokinetic data set was split into three parts (60:20:20) for training, validation, and testing by the improved maximum dissimilarity algorithm with the representative initial set selection algorithm and the weighted distance function. The multitask learning techniques enhanced the model predictive ability. The integrated transfer learning and multitask learning model demonstrated the best accuracies, because deep neural networks have the general feature extraction ability; transfer learning and multitask learning improve the model generalization.

CONCLUSIONS

The integrated transfer learning and multitask learning approach with the improved data set splitting algorithm was first introduced to predict the pharmacokinetic parameters. This method can be further employed in drug discovery and development.

Robust QSAR prediction models for volume of distribution at steady state in humans using relative distance measurements

The building of quantitative structure-activity relationship (QSAR) models for the in silico prediction of volume distribution for drugs at steady-state levels is vital for the selection of potential drugs at the synthesis stage. Using molecular descriptor matrixes, some regression models presenting low accuracy have been proposed, mainly due to the difficulty of compiling an appropriate dataset and the lack of information on dataset representation. In this paper, we use a benchmark dataset of very diverse drugs for the development of predictive models for volume distribution based on the use of relative distance matrixes as the input data to QSAR algorithms. Support vector machine, complex tree, bagged tree and Gaussian process regression algorithms were tested for fingerprint, similarity and relative distance matrixes used as input data, and the results of the built models were compared. Relative distance matrixes generated robust regression models in the training and external validation stages performed using cross-validation, obtaining values for correlation coefficient, bias, slope and root-mean-square error close to the ideal. Relative distance matrixes were also used for the design of classification models, obtaining excellent results with values of accuracy and area under receiver operating characteristic (AUC) close to 100%.

Automatically updating predictive modeling workflows support decision-making in drug design

Using predictive models for early decision-making in drug discovery has become standard practice. We suggest that model building needs to be automated with minimum input and low technical maintenance requirements. Models perform best when tailored to answering specific compound optimization related questions. If qualitative answers are required, 2-bin classification models are preferred. Integrating predictive modeling results with structural information stimulates better decision making. For in silico models supporting rapid structure–activity relationship cycles the performance deteriorates within weeks. Frequent automated updates of predictive models ensure best predictions. Consensus between multiple modeling approaches increases the prediction confidence. Combining qualified and nonqualified data optimally uses all available information. Dose predictions provide a holistic alternative to multiple individual property predictions for reaching complex decisions.

Estimation of elimination half-lives of organic chemicals in humans using gradient boosting machine

BACKGROUND

Elimination half-life is an important pharmacokinetic parameter that determines exposure duration to approach steady state of drugs and regulates drug administration. The experimental evaluation of half-life is time-consuming and costly. Thus, it is attractive to build an accurate prediction model for half-life.

METHODS

In this study, several machine learning methods, including gradient boosting machine (GBM), support vector regressions (RBF-SVR and Linear-SVR), local lazy regression (LLR), SA, SR, and GP, were employed to build high-quality prediction models. Two strategies of building consensus models were explored to improve the accuracy of prediction. Moreover, the applicability domains (ADs) of the models were determined by using the distance-based threshold.

RESULTS

Among seven individual models, GBM showed the best performance (R(2)=0.820 and RMSE=0.555 for the test set), and Linear-SVR produced the inferior prediction accuracy (R(2)=0.738 and RMSE=0.672). The use of distance-based ADs effectively determined the scope of QSAR models. However, the consensus models by combing the individual models could not improve the prediction performance. Some essential descriptors relevant to half-life were identified and analyzed.

CONCLUSIONS

An accurate prediction model for elimination half-life was built by GBM, which was superior to the reference model (R(2)=0.723 and RMSE=0.698).

GENERAL SIGNIFICANCE

Encouraged by the promising results, we expect that the GBM model for elimination half-life would have potential applications for the early pharmacokinetic evaluations, and provide guidance for designing drug candidates with favorable in vivo exposure profile. This article is part of a Special Issue entitled "System Genetics" Guest Editor: Dr. Yudong Cai and Dr. Tao Huang.

How to Choose In Vitro Systems to Predict In Vivo Drug Clearance: A System Pharmacology Perspective

The use of in vitro metabolism data to predict human clearance has become more significant in the current prediction of large scale drug clearance for all the drugs. The relevant information (in vitro metabolism data and in vivo human clearance values) of thirty-five drugs that satisfied the entry criteria of probe drugs was collated from the literature. Then the performance of different in vitro systems including Escherichia coli system, yeast system, lymphoblastoid system and baculovirus system is compared after in vitro-in vivo extrapolation. Baculovirus system, which can provide most of the data, has almost equal accuracy as the other systems in predicting clearance. And in most cases, baculovirus system has the smaller CV in scaling factors. Therefore, the baculovirus system can be recognized as the suitable system for the large scale drug clearance prediction.

Predicting volume of distribution with decision tree-based regression methods using predicted tissue:plasma partition coefficients

BACKGROUND

Volume of distribution is an important pharmacokinetic property that indicates the extent of a drug's distribution in the body tissues. This paper addresses the problem of how to estimate the apparent volume of distribution at steady state (Vss) of chemical compounds in the human body using decision tree-based regression methods from the area of data mining (or machine learning). Hence, the pros and cons of several different types of decision tree-based regression methods have been discussed. The regression methods predict Vss using, as predictive features, both the compounds' molecular descriptors and the compounds' tissue:plasma partition coefficients (Kt:p) - often used in physiologically-based pharmacokinetics. Therefore, this work has assessed whether the data mining-based prediction of Vss can be made more accurate by using as input not only the compounds' molecular descriptors but also (a subset of) their predicted Kt:p values.

RESULTS

Comparison of the models that used only molecular descriptors, in particular, the Bagging decision tree (mean fold error of 2.33), with those employing predicted Kt:p values in addition to the molecular descriptors, such as the Bagging decision tree using adipose Kt:p (mean fold error of 2.29), indicated that the use of predicted Kt:p values as descriptors may be beneficial for accurate prediction of Vss using decision trees if prior feature selection is applied.

CONCLUSIONS

Decision tree based models presented in this work have an accuracy that is reasonable and similar to the accuracy of reported Vss inter-species extrapolations in the literature. The estimation of Vss for new compounds in drug discovery will benefit from methods that are able to integrate large and varied sources of data and flexible non-linear data mining methods such as decision trees, which can produce interpretable models. Graphical AbstractDecision trees for the prediction of tissue partition coefficient and volume of distribution of drugs.

Applying machine learning techniques for ADME-Tox prediction: a review

INTRODUCTION

Pharmacokinetics involves the study of absorption, distribution, metabolism, excretion and toxicity of xenobiotics (ADME-Tox). In this sense, the ADME-Tox profile of a bioactive compound can impact its efficacy and safety. Moreover, efficacy and safety were considered some of the major causes of clinical failures in the development of new chemical entities. In this context, machine learning (ML) techniques have been often used in ADME-Tox studies due to the existence of compounds with known pharmacokinetic properties available for generating predictive models.

AREAS COVERED

This review examines the growth in the use of some ML techniques in ADME-Tox studies, in particular supervised and unsupervised techniques. Also, some critical points (e.g., size of the data set and type of output variable) must be considered during the generation of models that relate ADME-Tox properties and biological activity.

EXPERT OPINION

ML techniques have been successfully employed in pharmacokinetic studies, helping the complex process of designing new drug candidates from the use of reliable ML models. An application of this procedure would be the prediction of ADME-Tox properties from studies of quantitative structure-activity relationships or the discovery of new compounds from a virtual screening using filters based on results obtained from ML techniques.

The postmortem redistribution of iso-α-acids in postmortem specimens

Iso-α-acids (IAA) and reduced IAA can be used as beer-specific ingredient congeners to confirm beer consumption when detected in blood and other specimens using a UHPLC-MS/MS method. Recent analysis of postmortem casework demonstrated a high prevalence of beer consumption and the possibility of providing the source of alcohol in forensic casework. Research outlined in this manuscript has examined the degree to which the interval after death and quality of blood affects the concentration of IAA in postmortem cases. Postmortem whole blood and serum were analyzed in cases where natural or reduced IAA groups were detected. The trans-IAA, cis-IAA, and tetrahydro-IAA (TIAA) groups were subject to postmortem redistribution, although only weakly associated with the length of time from death to collection of specimens. Serum had threefold higher concentrations than blood for trans-IAA, cis-IAA, and TIAA. These studies confirm that although postmortem concentrations cannot be easily compared to concentrations found in living persons the presented findings do provide some understanding to assist in interpretation where the confirmation of beer consumption is required in forensic casework.

Applying linear and non-linear methods for parallel prediction of volume of distribution and fraction of unbound drug

Volume of distribution and fraction unbound are two key parameters in pharmacokinetics. The fraction unbound describes the portion of free drug in plasma that may extravasate, while volume of distribution describes the tissue access and binding of a drug. Reliable in silico predictions of these pharmacokinetic parameters would benefit the early stages of drug discovery, as experimental measuring is not feasible for screening purposes. We have applied linear and nonlinear multivariate approaches to predict these parameters: linear partial least square regression and non-linear recursive partitioning classification. The volume of distribution and fraction of unbound drug in plasma are predicted in parallel within the model, since the two are expected to be affected by similar physicochemical drug properties. Predictive models for both parameters were built and the performance of the linear models compared to models included in the commercial software Volsurf+. Our models performed better in predicting the unbound fraction (Q² 0.54 for test set compared to 0.38 with Volsurf+ model), but prediction accuracy of the volume of distribution was comparable to the Volsurf+ model (Q² of 0.70 for test set compared to 0.71 with Volsurf+ model). The nonlinear classification models were able to identify compounds with a high or low volume of distribution (sensitivity 0.81 and 0.71, respectively, for test set), while classification of fraction unbound was less successful. The interrelationship between the volume of distribution and fraction unbound is investigated and described in terms of physicochemical descriptors. Lipophilicity and solubility descriptors were found to have a high influence on both volume of distribution and fraction unbound, but with an inverse relationship.

PaDEL-DDPredictor: open-source software for PD-PK-T prediction

ADMET (absorption, distribution, metabolism, excretion, and toxicity)-related failure of drug candidates is a major issue for the pharmaceutical industry today. Prediction of PD-PK-T properties using in silico tools has become very important in pharmaceutical research to reduce cost and enhance efficiency. PaDEL-DDPredictor is an in silico tool for rapid prediction of PD-PK-T properties of compounds from their chemical structures. It is free and open-source software that, has both graphical user interface and command line interface, can work on all major platforms (Windows, Linux, and MacOS) and supports more than 90 different molecular file formats. The software can be downloaded from http://padel.nus.edu.sg/software/padelddpredictor.