Prediction of the tissue/blood partition coefficients of organic compounds based on the molecular structure using least-squares support vector machines

Machine learning and artificial intelligence in physiologically based pharmacokinetic modeling

Physiologically based pharmacokinetic (PBPK) models are useful tools in drug development and risk assessment of environmental chemicals. PBPK model development requires the collection of species-specific physiological, and chemical-specific absorption, distribution, metabolism, and excretion (ADME) parameters, which can be a time-consuming and expensive process. This raises a need to create computational models capable of predicting input parameter values for PBPK models, especially for new compounds. In this review, we summarize an emerging paradigm for integrating PBPK modeling with machine learning (ML) or artificial intelligence (AI)-based computational methods. This paradigm includes 3 steps (1) obtain time-concentration PK data and/or ADME parameters from publicly available databases, (2) develop ML/AI-based approaches to predict ADME parameters, and (3) incorporate the ML/AI models into PBPK models to predict PK summary statistics (eg, area under the curve and maximum plasma concentration). We also discuss a neural network architecture "neural ordinary differential equation (Neural-ODE)" that is capable of providing better predictive capabilities than other ML methods when used to directly predict time-series PK profiles. In order to support applications of ML/AI methods for PBPK model development, several challenges should be addressed (1) as more data become available, it is important to expand the training set by including the structural diversity of compounds to improve the prediction accuracy of ML/AI models; (2) due to the black box nature of many ML models, lack of sufficient interpretability is a limitation; (3) Neural-ODE has great potential to be used to generate time-series PK profiles for new compounds with limited ADME information, but its application remains to be explored. Despite existing challenges, ML/AI approaches will continue to facilitate the efficient development of robust PBPK models for a large number of chemicals.

A Computational Framework for Studying Gut-Brain Axis in Autism Spectrum Disorder

Introduction

The integrity of the intestinal epithelium is crucial for human health and is harmed in autism spectrum disorder (ASD). An aberrant gut microbial composition resulting in gut-derived metabolic toxins was found to damage the intestinal epithelium, jeopardizing tissue integrity. These toxins further reach the brain via the gut-brain axis, disrupting the normal function of the brain. A mechanistic understanding of metabolic disturbances in the brain and gut is essential to design effective therapeutics and early intervention to block disease progression. Herein, we present a novel computational framework integrating constraint based tissue specific metabolic (CBM) model and whole-body physiological pharmacokinetics (PBPK) modeling for ASD. Furthermore, the role of gut microbiota, diet, and oxidative stress is analyzed in ASD.

Methods

A representative gut model capturing host-bacteria and bacteria-bacteria interaction was developed using CBM techniques and patient data. Simultaneously, a PBPK model of toxin metabolism was assembled, incorporating multi-scale metabolic information. Furthermore, dynamic flux balance analysis was performed to integrate CBM and PBPK. The effectiveness of a probiotic and dietary intervention to improve autism symptoms was tested on the integrated model.

Results

The model accurately highlighted critical metabolic pathways of the gut and brain that are associated with ASD. These include central carbon, nucleotide, and vitamin metabolism in the host gut, and mitochondrial energy and amino acid metabolisms in the brain. The proposed dietary intervention revealed that a high-fiber diet is more effective than a western diet in reducing toxins produced inside the gut. The addition of probiotic bacteria Lactobacillus acidophilus, Bifidobacterium longum longum, Akkermansia muciniphila, and Prevotella ruminicola to the diet restores gut microbiota balance, thereby lowering oxidative stress in the gut and brain.

Conclusion

The proposed computational framework is novel in its applicability, as demonstrated by the determination of the whole-body distribution of ROS toxins and metabolic association in ASD. In addition, it emphasized the potential for developing novel therapeutic strategies to alleviate autism symptoms. Notably, the presented integrated model validates the importance of combining PBPK modeling with COBRA -specific tissue details for understanding disease pathogenesis.

Prediction of Terpenoid Toxicity Based on a Quantitative Structure-Activity Relationship Model

Terpenoids, including monoterpenoids (C10), norisoprenoids (C13), and sesquiterpenoids (C15), constitute a large group of plant-derived naturally occurring secondary metabolites with highly diverse chemical structures. A quantitative structure-activity relationship (QSAR) model to predict terpenoid toxicity and to evaluate the influence of their chemical structures was developed in this study by assessing in real time the toxicity of 27 terpenoid standards using the Gram-negative bioluminescent Vibrio fischeri. Under the test conditions, at a concentration of 1 µM, the terpenoids showed a toxicity level lower than 5%, with the exception of geraniol, citral, (S)-citronellal, geranic acid, (±)-α-terpinyl acetate, and geranyl acetone. Moreover, the standards tested displayed a toxicity level higher than 30% at concentrations of 50-100 µM, with the exception of (+)-valencene, eucalyptol, (+)-borneol, guaiazulene, β-caryophellene, and linalool oxide. Regarding the functional group, terpenoid toxicity was observed in the following order: alcohol > aldehyde ~ ketone > ester > hydrocarbons. The CODESSA software was employed to develop QSAR models based on the correlation of terpenoid toxicity and a pool of descriptors related to each chemical structure. The QSAR models, based on t-test values, showed that terpenoid toxicity was mainly attributed to geometric (e.g., asphericity) and electronic (e.g., maximum partial charge for a carbon (C) atom (Zefirov's partial charge (PC)) descriptors. Statistically, the most significant overall correlation was the four-parameter equation with a training coefficient and test coefficient correlation higher than 0.810 and 0.535, respectively, and a square coefficient of cross-validation (Q2) higher than 0.689. According to the obtained data, the QSAR models are suitable and rapid tools to predict terpenoid toxicity in a diversity of food products.

Hybrid connectionist model determines CO2-oil swelling factor

In-depth understanding of interactions between crude oil and CO₂ provides insight into the CO₂-based enhanced oil recovery (EOR) process design and simulation. When CO₂ contacts crude oil, the dissolution process takes place. This phenomenon results in the oil swelling, which depends on the temperature, pressure, and composition of the oil. The residual oil saturation in a CO₂-based EOR process is inversely proportional to the oil swelling factor. Hence, it is important to estimate this influential parameter with high precision. The current study suggests the predictive model based on the least-squares support vector machine (LS-SVM) to calculate the CO₂-oil swelling factor. A genetic algorithm is used to optimize hyperparameters (γ and σ ²) of the LS-SVM model. This model showed a high coefficient of determination (R ² = 0.9953) and a low value for the mean-squared error (MSE = 0.0003) based on the available experimental data while estimating the CO₂-oil swelling factor. It was found that LS-SVM is a straightforward and accurate method to determine the CO₂-oil swelling factor with negligible uncertainty. This method can be incorporated in commercial reservoir simulators to include the effect of the CO₂-oil swelling factor when adequate experimental data are not available.

Modeling bioconcentration factor (BCF) using mechanistically interpretable descriptors computed from open source tool "PaDEL-Descriptor"

Predictive regression-based models for bioconcentration factor (BCF) have been developed using mechanistically interpretable descriptors computed from open source tool PaDEL-Descriptor ( http://padel.nus.edu.sg/software/padeldescriptor/ ). A data set of 522 diverse chemicals has been used for this modeling study, and extended topochemical atom (ETA) indices developed by the present authors' group were chosen as the descriptors. Due to the importance of lipohilicity in modeling BCF, XLogP (computed partition coefficient) was also tried as an additional descriptor. Genetic function approximation followed by multiple linear regression algorithm was applied to select descriptors, and subsequent partial least squares analyses were performed to establish mathematical equations for BCF prediction. The model generated from only ETA indices shows importance of seven descriptors in model development, while the model generated from ETA descriptors along with XlogP shows importance of four descriptors in model development. In general, BCF depends on lipophilicity, presence of heteroatoms, presence of halogens, fused ring system, hydrogen bonding groups, etc. The developed models show excellent statistical qualities and predictive ability. The developed models were used also for prediction of an external data set available from the literature, and good quality of predictions (R (2) pred = 0.812 and 0.826) was demonstrated. Thus, BCF can be predicted using ETA and XlogP descriptors calculated from open source PaDEL-Descriptor software in the context of aquatic chemical toxicity management.

QSPR models for predicting log P(liver) values for volatile organic compounds combining statistical methods and domain knowledge

Volatile organic compounds (VOCs) are contained in a variety of chemicals that can be found in household products and may have undesirable effects on health. Thereby, it is important to model blood-to-liver partition coefficients (log P_liver) for VOCs in a fast and inexpensive way. In this paper, we present two new quantitative structure-property relationship (QSPR) models for the prediction of log P_liver, where we also propose a hybrid approach for the selection of the descriptors. This hybrid methodology combines a machine learning method with a manual selection based on expert knowledge. This allows obtaining a set of descriptors that is interpretable in physicochemical terms. Our regression models were trained using decision trees and neural networks and validated using an external test set. Results show high prediction accuracy compared to previous log P_liver models, and the descriptor selection approach provides a means to get a small set of descriptors that is in agreement with theoretical understanding of the target property.

Parameters for pyrethroid insecticide QSAR and PBPK/PD models for human risk assessment

In this review we have examined the status of parameters required by pyrethroid QSAR-PBPK/PD models for assessing health risks. In lieu of the chemical,biological, biochemical, and toxicological information developed on the pyrethroids since 1968, the finding of suitable parameters for QSAR and PBPK/PD model development was a monumental task. The most useful information obtained came from rat toxicokinetic studies (i.e., absorption, distribution, and excretion), metabolism studies with 14C-cyclopropane- and alcohol-labeled pyrethroids, the use of known chiral isomers in the metabolism studies and their relation to commercial products. In this review we identify the individual chiralisomers that have been used in published studies and the chiral HPLC columns available for separating them. Chiral HPLC columns are necessary for isomer identification and for developing kinetic values (Vm,, and Kin) for pyrethroid hydroxylation. Early investigators synthesized analytical standards for key pyrethroid metabolites, and these were used to confirm the identity of urinary etabolites, by using TLC. These analytical standards no longer exist, and muste resynthesized if further studies on the kinetics of the metabolism of pyrethroids are to be undertaken.In an attempt to circumvent the availability of analytical standards, several CYP450 studies were carried out using the substrate depletion method. This approach does not provide information on the products formed downstream, and may be of limited use in developing human environmental exposure PBPK/PD models that require extensive urinary metabolite data. Hydrolytic standards (i.e., alcohols and acids) were available to investigators who studied the carboxylesterase-catalyzed hydrolysis of several pyrethroid insecticides. The data generated in these studies are suitable for use in developing human exposure PBPK/PD models.Tissue:blood partition coefficients were developed for the parent pyrethroids and their metabolites, by using a published mechanistic model introduced by Poulin and Thiele (2002a; b) and log DpH 7.4 values. The estimated coefficients, especially those of adipose tissue, were too high and had to be corrected by using a procedure in which the proportion of parent or metabolite residues that are unbound to plasma albumin is considered, as described in the GastroPlus model (Simulations Plus, Inc.,Lancaster, CA). The literature suggested that Km values be adjusted by multiplying Km by the substrate (decimal amount) that is unbound to microsomal or CYPprotein. Mirfazaelian et al. (2006) used flow- and diffusion-limited compartments in their deltamethrin model. The addition of permeability areas (PA) having diffusion limits, such as the fat and slowly perfused compartments, enabled the investigators to bring model predictions in line with in vivo data.There appears to be large differences in the manner and rate of absorption of the pyrethroids from the gastrointestinal tract, implying that GI advanced compartmental transit models (ACAT) need to be included in PBPK models. This is especially true of the absorption of an oral dose of tefluthrin in male rats, in which 3.0-6.9%,41.3-46.3%, and 5.2-15.5% of the dose is eliminated in urine, feces, and bile,respectively (0-48 h after administration). Several percutaneous studies with the pyrethroids strongly support the belief that these insecticides are not readily absorbed, but remain on the surface of the skin until they are washed off. In one articular study (Sidon et al. 1988) the high levels of permethrin absorption through the forehead skin (24-28%) of the monkey was reported over a 7- to 14-days period.Wester et al. (1994) reported an absorption of 1.9% of pyrethrin that had been applied to the forearm of human volunteers over a 7-days period.SAR models capable of predicting the binding of the pyrethroids to plasma and hepatic proteins were developed by Yamazaki and Kanaoka (2004), Saiakhov et al. (2000), Colmenarejo et al. (2001), and Colmenarejo (2003). QikProp(Schrodinger, LLC) was used to obtain Fu values for calculating partition coefficients and for calculating permeation constants (Caco-2, MDCK, and logBBB). ADMET Predictor (Simulations Plus Inc.) provided Vm~,x and Km values for the hydroxylation of drugs/pyrethroids by human liver recombinant cytochrome P450 enzymes making the values available for possible use in PBPK/PD models.The Caco-2 permeability constants and CYP3A4 Vmax and Km values are needed in PBPK/PD models with GI ACAT sub models. Modeling work by Chang et al.(2009) produced rate constants (kcat) for the hydrolysis of pyrethroids by rat serumcarboxylesterases. The skin permeation model of Potts and Guy (1992) was used topredict K, values for the dermal absorption of the 15 pyrethroids.The electrophysiological studies by Narahashi (1971) and others (Breckenridgeet al. 2009; Shafer et al. 2005; Soderlund et al. 2002; Wolansky and Harrill 2008)demonstrated that the mode of action of pyrethroids on nerves is to interfere with the changes in sodium and potassium ion currents. The pyrethroids, being highly lipid soluble, are bound or distributed in lipid bilayers of the nerve cell membrane and exert their action on sodium channel proteins. The rising phase of the action potential is caused by sodium influx (sodium activation), while the falling phase is caused by sodium activation being turned off, and an increase in potassium efflux(potassium activation). The action of allethrin and other pyrethroids is caused by an inhibition or block of the normal currents. An equation by Tatebayashi and Narahashi (1994) that describes the action of pyrethroids on sodium channels was found in the literature. This equation, or some variation of it, may be suitable for use in the PD portion of pyrethroid PBPK models.

The use of machine learning and nonlinear statistical tools for ADME prediction

Absorption, distribution, metabolism and excretion (ADME)-related failure of drug candidates is a major issue for the pharmaceutical industry today. Prediction of ADME by in silico tools has now become an inevitable paradigm to reduce cost and enhance efficiency in pharmaceutical research. Recently, machine learning as well as nonlinear statistical tools has been widely applied to predict routine ADME end points. To achieve accurate and reliable predictions, it would be a prerequisite to understand the concepts, mechanisms and limitations of these tools. Here, we have devised a small synthetic nonlinear data set to help understand the mechanism of machine learning by 2D-visualisation. We applied six new machine learning methods to four different data sets. The methods include Naive Bayes classifier, classification and regression tree, random forest, Gaussian process, support vector machine and k nearest neighbour. The results demonstrated that ensemble learning and kernel machine displayed greater accuracy of prediction than classical methods irrespective of the data set size. The importance of interaction with the engineering field is also addressed. The results described here provide insights into the mechanism of machine learning, which will enable appropriate usage in the future.

Current mathematical methods used in QSAR/QSPR studies

This paper gives an overview of the mathematical methods currently used in quantitative structure-activity/property relationship (QASR/QSPR) studies. Recently, the mathematical methods applied to the regression of QASR/QSPR models are developing very fast, and new methods, such as Gene Expression Programming (GEP), Project Pursuit Regression (PPR) and Local Lazy Regression (LLR) have appeared on the QASR/QSPR stage. At the same time, the earlier methods, including Multiple Linear Regression (MLR), Partial Least Squares (PLS), Neural Networks (NN), Support Vector Machine (SVM) and so on, are being upgraded to improve their performance in QASR/QSPR studies. These new and upgraded methods and algorithms are described in detail, and their advantages and disadvantages are evaluated and discussed, to show their application potential in QASR/QSPR studies in the future.

Biopartitioning micellar chromatography to predict blood to lung, blood to liver, blood to fat and blood to skin partition coefficients of drugs

Biopartitioning micellar chromatography (BMC), a mode of micellar liquid chromatography that uses micellar mobile phases of Brij35 in adequate experimental conditions, has demonstrated to be useful in mimicking the drug partitioning process into biological systems. In this paper, the usefulness of BMC for predicting the partition coefficients from blood to lung, blood to liver, blood to fat and blood to skin is demonstrated. PLS2 and multiple linear regression (MLR) models based on BMC retention data are proposed and compared with other ones reported in bibliography. The proposed models present better or similar descriptive and predictive capability.

Parameters for Carbamate Pesticide QSAR and PBPK/PD Models for Human Risk Assessment

Our interest in providing parameters for the development of quantitative structure physiologically based pharmacokinetic/pharmacodynamic (QSPBPK/PD) models for assessing health risks to carbamates (USEPA 2005) comes from earlier work with organophosphorus (OP) insecticides (Knaak et al. 2004). Parameters specific to each carbamate are needed in the construction of PBPK/PD models along with their metabolic pathways. Parameters may be obtained by (1) development of QSAR models, (2) collecting pharmacokinetic data, and (3) determining pharmacokinetic parameters by fitting to experimental data. The biological parameters are given in Table 1 (Blancato et al. 2000). Table 1 Biological Parameters Required for Carbamate Pesticide Physiologically Based Pharmacokinetic/Pharmacodynamic (PBPK/PD) Models.(a).

Quantitative structure-pharmacokinetic/pharmacodynamic relationships

Quantitative structure-activity relationships have long been considered a vital component of drug discovery and development, providing insight into the role of molecular properties in the biological activity of similar and unrelated compounds. Recognition that in vitro bioassay and/or pre-clinical activity are insufficient for anticipating which compounds are suitable leads for further development has shifted the focus toward integrated pharmacokinetic (PK) and pharmacodynamic (PD) processes. Over the last decade, considerable progress has been made in constructing empirical and mechanistic quantitative structure-PK relationships (QSPKR), as well as diverse mechanism-based pharmacodynamic models of drug effects. In this review, traditional and contemporary approaches to developing QSPKR models are discussed, along with selected examples of attempts to couple QSPKR and pharmacodynamic models to anticipate the intensity and time-course of the pharmacological effects of new or related compounds, or quantitative structure-pharmacodynamic relationships modeling. Such models are in accordance with the goals of systems biology and the ideal of designing drugs and delivery systems from first principles.