In silico ADME modelling: prediction models for blood-brain barrier permeation using a systematic variable selection method

Exploring blood-brain barrier passage using atomic weighted vector and machine learning

CONTEXT

This study investigates the potential of leveraging molecular properties, as determined by MD-LOVIs software and machine learning techniques, to predict the ability of compounds to cross the blood-brain barrier (BBB). Accurate prediction of BBB permeation is critical for the development of central nervous system (CNS) drugs. The study applies various machine learning models, including both classification and regression techniques, to predict BBB passage and molecular activity. Notably, classification models such as GBM-AWV (accuracy = 0.801), GLM-CN (accuracy = 0.808), SVMPoly-means (accuracy = 0.980), SVMPoly-AC (accuracy = 0.980), SVMPoly-MI_TI_SI (accuracy = 0.900), SVMPoly-GI (accuracy = 0.900), RF-means (accuracy = 0.870), and GLM-means (accuracy = 0.818) demonstrate high accuracy in predicting BBB passage. In contrast, regression models like ES-RLM-AG (R2 = 0.902), IB-IBK (R2 = 0.82), IB-Kstar (R2 = 0.834), IB-MLP (R2 = 0.843), and DRF-AWV (R2 = 0.810) exhibit strong performance in predicting molecular activity. The results show that classification models like GBM-AWV, GLM-CN, and SVMPoly variants, as well as regression models like ES-RLM-AG and IB-MLP, achieve high performance, demonstrating the effectiveness of machine learning in predicting BBB permeability.

METHODS

The computational methods employed in this study include the MD-LOVIs software for generating molecular descriptors and several machine learning algorithms, including gradient boosting machines (GBM), generalized linear models (GLM), support vector machines (SVM) with polynomial kernels, random forests (RF), ensemble regression models, and instance-based learning algorithms. These models were trained and validated using various datasets to predict BBB passage and molecular activity, with the performance metrics reported for each model. Standard computational techniques were employed throughout, ensuring the reliability of the predictions.

Development of QSAR models to predict blood-brain barrier permeability

Assessing drug permeability across the blood-brain barrier (BBB) is important when evaluating the abuse potential of new pharmaceuticals as well as developing novel therapeutics that target central nervous system disorders. One of the gold-standard in vivo methods for determining BBB permeability is rodent log BB; however, like most in vivo methods, it is time-consuming and expensive. In the present study, two statistical-based quantitative structure-activity relationship (QSAR) models were developed to predict BBB permeability of drugs based on their chemical structure. The in vivo BBB permeability data were harvested for 921 compounds from publicly available literature, non-proprietary drug approval packages, and University of Washington's Drug Interaction Database. The cross-validation performance statistics for the BBB models ranged from 82 to 85% in sensitivity and 80-83% in negative predictivity. Additionally, the performance of newly developed models was assessed using an external validation set comprised of 83 chemicals. Overall, performance of individual models ranged from 70 to 75% in sensitivity, 70-72% in negative predictivity, and 78-86% in coverage. The predictive performance was further improved to 93% in coverage by combining predictions across the two software programs. These new models can be rapidly deployed to predict blood brain barrier permeability of pharmaceutical candidates and reduce the use of experimental animals.

Biomimetic Chromatographic Studies Combined with the Computational Approach to Investigate the Ability of Triterpenoid Saponins of Plant Origin to Cross the Blood-Brain Barrier

Biomimetic (non-cell based in vitro) and computational (in silico) studies are commonly used as screening tests in laboratory practice in the first stages of an experiment on biologically active compounds (potential drugs) and constitute an important step in the research on the drug design process. The main aim of this study was to evaluate the ability of triterpenoid saponins of plant origin to cross the blood-brain barrier (BBB) using both computational methods, including QSAR methodology, and biomimetic chromatographic methods, i.e., High Performance Liquid Chromatography (HPLC) with Immobilized Artificial Membrane (IAM) and cholesterol (CHOL) stationary phases, as well as Bio-partitioning Micellar Chromatography (BMC). The tested compounds were as follows: arjunic acid (Terminalia arjuna), akebia saponin D (Akebia quinata), bacoside A (Bacopa monnieri) and platycodin D (Platycodon grandiflorum). The pharmacokinetic BBB parameters calculated in silico show that three of the four substances, i.e., arjunic acid, akebia saponin D, and bacoside A exhibit similar values of brain/plasma equilibration rate expressed as logPSFubrain (the average logPSFubrain: -5.03), whereas the logPSFubrain value for platycodin D is -9.0. Platycodin D also shows the highest value of the unbound fraction in the brain obtained using the examined compounds (0.98). In these studies, it was found out for the first time that the logarithm of the analyte-micelle association constant (logKMA) calculated based on Foley's equation can describe the passage of substances through the BBB. The most similar logBB values were obtained for hydrophilic platycodin D, applying both biomimetic and computational methods. All of the obtained logBB values and physicochemical parameters of the molecule indicate that platycodin D does not cross the BBB (the average logBB: -1.681), even though the in silico estimated value of the fraction unbound in plasma is relatively high (0.52). As far as it is known, this is the first paper that shows the applicability of biomimetic chromatographic methods in predicting the penetration of triterpenoid saponins through the BBB.

Nonlinear Method to Predict the Distribution of Structurally Diverse Compounds between Blood and Tissue

In this work, methods for predicting the distribution of compounds between blood and tissue were investigated using nonlinear regression analysis. For the tissue/blood partition coefficient, 282 compounds and 810 activity data for seven tissues were selected. Twenty-four parameters were studied for each state of the compound. A study set with the most compounds and activity data in similar studies was established, and a model with good prediction ability was obtained. A total of 773 data points were randomly divided into two data sets: the training set contained 623 data points (n = 623, r = 0.822, s = 0.438, F = 142.2, and Q = 0.814) and the test set contained 150 data points (n = 150, r = 0.814, and s = 0.334). Furthermore, individual tissue/blood distribution coefficients were also studied in depth to obtain separate models for predicting the distribution coefficient of a drug between blood and a tissue. By applying these models, not only can the tissue/blood distribution coefficient or single tissue/blood distribution coefficient of the seven tissues and organs of the human body be predicted, but the distribution of the drug molecules in different states (neutral, cation, and anion) in the three tissue components can also be predicted.

In Silico Studies on Triterpenoid Saponins Permeation through the Blood-Brain Barrier Combined with Postmortem Research on the Brain Tissues of Mice Affected by Astragaloside IV Administration

As the number of central nervous system (CNS) drug candidates is constantly growing, there is a strong need for precise a priori prediction of whether an administered compound is able to cross the blood–brain barrier (BBB). The aim of this study was to evaluate the ability to cross the BBB of triterpenoid saponins occurring in Astragalus mongholicus roots. The research was carried out using in silico methods combined with postmortem studies on the brain tissues of mice treated with isolated astragaloside IV (AIV). Firstly, to estimate the ability to cross the BBB by the tested saponins, new quantitative structure–activity relationship (QSAR) models were established. The reliability and predictability of the model based on the values of the blood–brain barrier penetration descriptor (logBB), the difference between the n-octanol/water and cyclohexane/water logP (ΔlogP), the logarithm of n-octanol/water partition coefficient (logP_ow), and the excess molar refraction (E) were both confirmed using the applicability domain (AD). The critical leverage value h* was found to be 0.128. The relationships between the standardized residuals and the leverages were investigated here. The application of an in vitro acetylcholinesterase-inhibition test showed that AIV can be recognized as the strongest inhibitor among the tested compounds. Therefore, it was isolated for the postmortem studies on brain tissues and blood using semi-preparative HPLC with the mobile phase composed of water, methanol, and ethyl acetate (1.7:2.1:16.2 v/v/v). The results of the postmortem studies on the brain tissues show a regular dependence of the final concentration of AIV in the analyzed brain samples of animals treated with 12.5 and 25 mg/kg b.w. of AIV (0.00012299 and 0.0002306 mg, respectively, per one brain). Moreover, the AIV logBB value was experimentally determined and found to be equal to 0.49 ± 0.03.

Barrier Properties and Transcriptome Expression in Human iPSC-Derived Models of the Blood-Brain Barrier

Cell-based models of the blood-brain barrier (BBB) are important for increasing the knowledge of BBB formation, degradation and brain exposure of drug substances. Human models are preferred over animal models because of interspecies differences in BBB structure and function. However, access to human primary BBB tissue is limited and has shown degeneration of BBB functions in vitro. Human induced pluripotent stem cells (iPSCs) can be used to generate relevant cell types to model the BBB with human tissue. We generated a human iPSC-derived model of the BBB that includes endothelial cells in coculture with pericytes, astrocytes and neurons. Evaluation of barrier properties showed that the endothelial cells in our coculture model have high transendothelial electrical resistance, functional efflux and ability to discriminate between CNS permeable and non-permeable substances. Whole genome expression profiling revealed transcriptional changes that occur in coculture, including upregulation of tight junction proteins, such as claudins and neurotransmitter transporters. Pathway analysis implicated changes in the WNT, TNF, and PI3K-Akt pathways upon coculture. Our data suggest that coculture of iPSC-derived endothelial cells promotes barrier formation on a functional and transcriptional level. The information about gene expression changes in coculture can be used to further improve iPSC-derived BBB models through selective pathway manipulation. Stem Cells 2018;36:1816-12.

Isolation of Cerebral Capillaries from Fresh Human Brain Tissue

Understanding blood-brain barrier function under physiological and pathophysiological conditions is critical for the development of new therapeutic strategies that hold the promise to enhance brain drug delivery, improve brain protection, and treat brain disorders. However, studying the human blood-brain barrier function is challenging. Thus, there is a critical need for appropriate models. In this regard, brain capillaries isolated from human brain tissue represent a unique tool to study barrier function as close to the human in vivo situation as possible. Here, we describe an optimized protocol to isolate capillaries from human brain tissue at a high yield and with consistent quality and purity. Capillaries are isolated from fresh human brain tissue using mechanical homogenization, density-gradient centrifugation, and filtration. After the isolation, the human brain capillaries can be used for various applications including leakage assays, live cell imaging, and immune-based assays to study protein expression and function, enzyme activity, or intracellular signaling. Isolated human brain capillaries are a unique model to elucidate the regulation of the human blood-brain barrier function. This model can provide insights into central nervous system (CNS) pathogenesis, which will help the development of therapeutic strategies for treating CNS disorders.

ADME properties evaluation in drug discovery: in silico prediction of blood-brain partitioning

The absorption, distribution, metabolism and excretion properties are important for drugs, and prediction of these properties in advance will save the cost of drug discovery substantially. The ability to penetrate the blood-brain barrier is critical for drugs targeting central nervous system, which is represented by the ratio of its concentration in brain and in blood. Herein, a quantitative structure-property relationship study was carried out to predict blood-brain partitioning coefficient (logBB) of a data set consisting of 287 compounds. Four different methods including support vector machine, multivariate linear regression, multivariate adaptive regression splines and random forest were employed to build prediction models with 116 molecular descriptors selected by Boruta algorithm. The RF model had best performance in training set ([Formula: see text] = 0.938), test set ([Formula: see text] = 0.840) and tenfold cross-validation ([Formula: see text] = 0.788). Finally, we found that the polar surface area and octanol-water partition coefficient have the greatest influence on blood-brain partitioning. Results suggest that the proposed model is a useful and practical tool to predict the logBB values of drug candidates.

The benefits of in silico modeling to identify possible small-molecule drugs and their off-target interactions

The research into the use of small molecules as drugs continues to be a key driver in the development of molecular databases, computer-aided drug design software and collaborative platforms. The evolution of computational approaches is driven by the essential criteria that a drug molecule has to fulfill, from the affinity to targets to minimal side effects while having adequate absorption, distribution, metabolism, and excretion (ADME) properties. A combination of ligand- and structure-based drug development approaches is already used to obtain consensus predictions of small molecule activities and their off-target interactions. Further integration of these methods into easy-to-use workflows informed by systems biology could realize the full potential of available data in the drug discovery and reduce the attrition of drug candidates.

A method to predict different mechanisms for blood-brain barrier permeability of CNS activity compounds in Chinese herbs using support vector machine

The blood-brain barrier (BBB), a highly selective barrier between central nervous system (CNS) and the blood stream, restricts and regulates the penetration of compounds from the blood into the brain. Drugs that affect the CNS interact with the BBB prior to their target site, so the prediction research on BBB permeability is a fundamental and significant research direction in neuropharmacology. In this study, we combed through the available data and then with the help of support vector machine (SVM), we established an experiment process for discovering potential CNS compounds and investigating the mechanisms of BBB permeability of them to advance the research in this field four types of prediction models, referring to CNS activity, BBB permeability, passive diffusion and efflux transport, were obtained in the experiment process. The first two models were used to discover compounds which may have CNS activity and also cross the BBB at the same time; the latter two were used to elucidate the mechanism of BBB permeability of those compounds. Three optimization parameter methods, Grid Search, Genetic Algorithm (GA), and Particle Swarm Optimization (PSO), were used to optimize the SVM models. Then, four optimal models were selected with excellent evaluation indexes (the accuracy, sensitivity and specificity of each model were all above 85%). Furthermore, discrimination models were utilized to study the BBB properties of the known CNS activity compounds in Chinese herbs and this may guide the CNS drug development. With the relatively systematic and quick approach, the application rationality of traditional Chinese medicines for treating nervous system disease in the clinical practice will be improved.

Towards Better BBB Passage Prediction Using an Extensive and Curated Data Set

In the present report, the challenging task of drug delivery across the blood-brain barrier (BBB) is addressed via a computational approach. The BBB passage was modeled using classification and regression schemes on a novel extensive and curated data set (the largest to the best of our knowledge) in terms of log BB. Prior to the model development, steps of data analysis that comprise chemical data curation, structural, cutoff and cluster analysis (CA) were conducted. Linear Discriminant Analysis (LDA) and Multiple Linear Regression (MLR) were used to fit classification and correlation functions. The best LDA-based model showed overall accuracies over 85 % and 83 % for the training and test sets, respectively. Also a MLR-based model with acceptable explanation of more than 69 % of the variance in the experimental log BB was developed. A brief and general interpretation of proposed models allowed the estimation on how 'near' our computational approach is to the factors that determine the passage of molecules through the BBB. In a final effort some popular and powerful Machine Learning methods were considered. Comparable or similar performance was observed respect to the simpler linear techniques. Most of the compounds with anomalous behavior were put aside into a set denoted as controversial set and discussion regarding to these compounds is provided. Finally, our results were compared with methodologies previously reported in the literature showing comparable to better results. The results could represent useful tools available and reproducible by all scientific community in the early stages of neuropharmaceutical drug discovery/development projects.

In vitro cerebrovascular modeling in the 21st century: current and prospective technologies

The blood-brain barrier (BBB) maintains the brain homeostasis and dynamically responds to events associated with systemic and/or rheological impairments (e.g., inflammation, ischemia) including the exposure to harmful xenobiotics. Thus, understanding the BBB physiology is crucial for the resolution of major central nervous system CNS) disorders challenging both health care providers and the pharmaceutical industry. These challenges include drug delivery to the brain, neurological disorders, toxicological studies, and biodefense. Studies aimed at advancing our understanding of CNS diseases and promoting the development of more effective therapeutics are primarily performed in laboratory animals. However, there are major hindering factors inherent to in vivo studies such as cost, limited throughput and translational significance to humans. These factors promoted the development of alternative in vitro strategies for studying the physiology and pathophysiology of the BBB in relation to brain disorders as well as screening tools to aid in the development of novel CNS drugs. Herein, we provide a detailed review including pros and cons of current and prospective technologies for modelling the BBB in vitro including ex situ, cell based and computational (in silico) models. A special section is dedicated to microfluidic systems including micro-BBB, BBB-on-a-chip, Neurovascular Unit-on-a-Chip and Synthetic Microvasculature Blood-brain Barrier.

In-silico prediction of blood-brain barrier permeability

The ability of penetration of the blood-brain barrier is one of the significant properties of a drug or drug-like compound for the central nervous system (CNS), which is commonly expressed by log BB (log BB = log (C (brain)/C (blood))). In this work, a dataset of 320 compounds with log BB values was split into a training set including 198 compounds and a test set including 122 compounds according to their structure properties by a Kohonen's self-organizing map (SOM). Each molecule was represented by global and shape descriptors, 2D autocorrelation descriptors and RDF descriptors calculated by ADRIANA.Code. Several quantitative models for prediction of log BB were built by a multilinear regression (MLR), a support vector machine (SVM) and an artificial neural network (ANN) analysis. The models show good prediction performance on the test set compounds.

Integrated pharmacokinetic-driven approach to screen candidate anticancer drugs for brain tumor chemotherapy

The goal of the study was to develop an effective screening strategy to select new agents for brain tumor chemotherapy from a series of low molecular weight anticancer agents [ON123x] by the combined use of in silico, in vitro cytotoxicity, and in vitro ADME profiling studies. The results of these studies were cast into a pipeline of tier 1 and tier 2 procedures that resulted in the identification of ON123300 as the lead compound. Of the 154 ON123xx compounds, 13 met tier 1 screening criteria based on physicochemical properties [i.e., MW < 450 Da, predicted log P between 2 and 3.5] and in vitro glioma cell cytotoxicity [i.e., IC50 < 10 μM] and were further tested in tier 2 assays. The tier 2 profiling studies consisted of metabolic stability, MDCK-MDR1 cell permeability and plasma and brain protein binding that were combined to globally assess whether favorable pharmacokinetic properties and brain penetration could be achieved in vivo. In vivo cassette dosing studies were conducted in mice for 12 compounds that permitted examination of in vitro/in vivo relationships that confirmed the suitability of the in vitro assays. A parameter derived from the in vitro assays accurately predicted the extent of drug accumulation in the brain based on the area under the drug concentration-time curve in brain measured in the cassette dosing study (r (2) = 0.920). Overall, the current studies demonstrated the value of an integrated pharmacokinetic-driven approach to identify potentially efficacious agents for brain tumor chemotherapy.

Qualitative prediction of blood-brain barrier permeability on a large and refined dataset

The prediction of blood-brain barrier permeation is vitally important for the optimization of drugs targeting the central nervous system as well as for avoiding side effects of peripheral drugs. Following a previously proposed model on blood-brain barrier penetration, we calculated the cross-sectional area perpendicular to the amphiphilic axis. We obtained a high correlation between calculated and experimental cross-sectional area (r = 0.898, n = 32). Based on these results, we examined a correlation of the calculated cross-sectional area with blood-brain barrier penetration given by logBB values. We combined various literature data sets to form a large-scale logBB dataset with 362 experimental logBB values. Quantitative models were calculated using bootstrap validated multiple linear regression. Qualitative models were built by a bootstrapped random forest algorithm. Both methods found similar descriptors such as polar surface area, pKa, logP, charges and number of positive ionisable groups to be predictive for logBB. In contrast to our initial assumption, we were not able to obtain models with the cross-sectional area chosen as relevant parameter for both approaches. Comparing those two different techniques, qualitative random forest models are better suited for blood-brain barrier permeability prediction, especially when reducing the number of descriptors and using a large dataset. A random forest prediction system (n(trees) = 5) based on only four descriptors yields a validated accuracy of 88%.

Identification of novel functional inhibitors of acid sphingomyelinase

We describe a hitherto unknown feature for 27 small drug-like molecules, namely functional inhibition of acid sphingomyelinase (ASM). These entities named FIASMAs (Functional Inhibitors of Acid SphingoMyelinAse), therefore, can be potentially used to treat diseases associated with enhanced activity of ASM, such as Alzheimer's disease, major depression, radiation- and chemotherapy-induced apoptosis and endotoxic shock syndrome. Residual activity of ASM measured in the presence of 10 µM drug concentration shows a bimodal distribution; thus the tested drugs can be classified into two groups with lower and higher inhibitory activity. All FIASMAs share distinct physicochemical properties in showing lipophilic and weakly basic properties. Hierarchical clustering of Tanimoto coefficients revealed that FIASMAs occur among drugs of various chemical scaffolds. Moreover, FIASMAs more frequently violate Lipinski's Rule-of-Five than compounds without effect on ASM. Inhibition of ASM appears to be associated with good permeability across the blood-brain barrier. In the present investigation, we developed a novel structure-property-activity relationship by using a random forest-based binary classification learner. Virtual screening revealed that only six out of 768 (0.78%) compounds of natural products functionally inhibit ASM, whereas this inhibitory activity occurs in 135 out of 2028 (6.66%) drugs licensed for medical use in humans.

Predictivity approach for quantitative structure-property models. Application for blood-brain barrier permeation of diverse drug-like compounds

The goal of the present research was to present a predictivity statistical approach applied on structure-based prediction models. The approach was applied to the domain of blood-brain barrier (BBB) permeation of diverse drug-like compounds. For this purpose, 15 statistical parameters and associated 95% confidence intervals computed on a 2 × 2 contingency table were defined as measures of predictivity for binary quantitative structure-property models. The predictivity approach was applied on a set of compounds comprised of 437 diverse molecules, 122 with measured BBB permeability and 315 classified as active or inactive. A training set of 81 compounds (~2/3 of 122 compounds assigned randomly) was used to identify the model and a test set of 41 compounds was used as the internal validation set. The molecular descriptor family on vertices cutting was the computation tool used to generate and calculate structural descriptors for all compounds. The identified model was assessed using the predictivity approach and compared to one model previously reported. The best-identified classification model proved to have an accuracy of 69% in the training set (95%CI [58.53–78.37]) and of 73% in the test set (95%CI [58.32–84.77]). The predictive accuracy obtained on the external set proved to be of 73% (95%CI [67.58–77.39]). The classification model proved to have better abilities in the classification of inactive compounds (specificity of ~74% [59.20–85.15]) compared to abilities in the classification of active compounds (sensitivity of ~64% [48.47–77.70]) in the training and external sets. The overall accuracy of the previously reported model seems not to be statistically significantly better compared to the identified model (~81% [71.45–87.80] in the training set, ~93% [78.12–98.17] in the test set and ~79% [70.19–86.58] in the external set). In conclusion, our predictivity approach allowed us to characterize the model obtained on the investigated set of compounds as well as compare it with a previously reported model. According to the obtained results, the reported model should be chosen if a correct classification of inactive compounds is desired and the previously reported model should be chosen if a correct classification of active compounds is most wanted.

Development of novel in silico model to predict corneal permeability for congeneric drugs: a QSPR approach

This study was undertaken to determine in vivo permeability coefficients for fluoroquinolones and to assess its correlation with the permeability derived using reported models in the literature. Further, the aim was to develop novel QSPR model to predict corneal permeability for fluoroquinolones and test its suitability on other training sets. The in vivo permeability coefficient was determined using cassette dosing (N-in-One) approach for nine fluoroquinolones (norfloxacin, ciprofloxacin, lomefloxacin, ofloxacin, levofloxacin, sparfloxacin, pefloxacin, gatifloxacin, and moxifloxacin) in rabbits. The correlation between corneal permeability derived using in vivo studies with that derived from reported models was determined. Novel QSPR-based model was developed using in vivo corneal permeability along with other molecular descriptors. The suitability of developed model was tested on β-blockers (n = 15). The model showed better prediction of corneal permeability for fluoroquinolones (r(2) > 0.9) as well as β-blockers (r(2) > 0.6). The newly developed QSPR model based upon in vivo generated data was found suitable to predict corneal permeability for fluoroquinolones as well as other sets of compounds.

A 2D-QSPR approach to predict blood-brain barrier penetration of drugs acting on the central nervous system

Drugs acting on the central nervous system (CNS) have to cross the blood-brain barrier (BBB) in order to perform their pharmacological actions. Passive BBB diffusion can be partially expressed by the blood/brain partition coefficient (logBB). As the experimental evaluation of logBB is time and cost consuming, theoretical methods such as quantitative structure-property relationships (QSPR) can be useful to predict logBB values. In this study, a 2D-QSPR approach was applied to a set of 28 drugs acting on the CNS, using the logBB property as biological data. The best QSPR model [n = 21, r = 0.94 (r² = 0.88), s = 0.28, and Q² = 0.82] presented three molecular descriptors: calculated n-octanol/water partition coefficient (ClogP), polar surface area (PSA), and polarizability (α). Six out of the seven compounds from the test set were well predicted, which corresponds to good external predictability (85.7%). These findings can be helpful to guide future approaches regarding those molecular descriptors which must be considered for estimating the logBB property, and also for predicting the BBB crossing ability for molecules structurally related to the investigated set.

Prediction of blood-brain partitioning: a model based on molecular electronegativity distance vector descriptors

The objective of this paper is to build a reliable model based on the molecular electronegativity distance vector (MEDV) descriptors for predicting the blood-brain barrier (BBB) permeability and to reveal the effects of the molecular structural segments on the BBB permeability. Using 70 structurally diverse compounds, the partial least squares regression (PLSR) models between the BBB permeability and the MEDV descriptors were developed and validated by the variable selection and modeling based on prediction (VSMP) technique. The estimation ability, stability, and predictive power of a model are evaluated by the estimated correlation coefficient (r), leave-one-out (LOO) cross-validation correlation coefficient (q), and predictive correlation coefficient (R(p)). It has been found that PLSR model has good quality, r=0.9202, q=0.7956, and R(p)=0.6649 for M1 model based on the training set of 57 samples. To search the most important structural factors affecting the BBB permeability of compounds, we performed the values of the variable importance in projection (VIP) analysis for MEDV descriptors. It was found that some structural fragments in compounds, such as -CH(3), -CH(2)-, =CH-, =C, triple bond C-, -CH<, =C<, =N-, -NH-, =O, and -OH, are the most important factors affecting the BBB permeability.

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.

Prediction of skin sensitization potential using D-optimal design and GA-kNN classification methods

Modelling of skin sensitization data of 255 diverse compounds and 450 calculated descriptors was performed to develop global predictive classification models that are applicable to whole chemical space. With this aim, we employed two automated procedures, (a) D-optimal design to select optimal members of the training and test sets and (b) k-Nearest Neighbour classification (kNN) method along with Genetic Algorithms (GA-kNN Classification) to select significant and independent descriptors in order to build the models. This methodology helped us to derive multiple models, M1-M5, that are stable and robust. The best among them, model M1 (CCR(train) = 84.3%, CCR(test) = 87.2% and CCR(ext) = 80.4%), is based on six neighbours and nine descriptors and further suggests that: (a) it is stable and robust and performs better than the reported models in literature, and (b) the combination of D-optimal design and GA-kNN classification approach is a very promising approach. Consensus prediction based on the models M1-M5 improved the CCR of training, test and external validation datasets by 3.8%, 4.45% and 3.85%, respectively, over M1. From the analysis of the physical meaning of the selected descriptors, it is inferred that the skin sensitization potential of small organic compounds can be accurately predicted using calculated descriptors that code for the following fundamental properties: (i) lipophilicity, (ii) atomic polarizability, (iii) shape, (iii) electrostatic interactions, and (iv) chemical reactivity.

QSAR Studies on andrographolide derivatives as α-glucosidase inhibitors

Andrographolide derivatives were shown to inhibit α-glucosidase. To investigate the relationship between activities and structures of andrographolide derivatives, a training set was chosen from 25 andrographolide derivatives by the principal component analysis (PCA) method, and a quantitative structure-activity relationship (QSAR) was established by 2D and 3D QSAR methods. The cross-validation r² (0.731) and standard error (0.225) illustrated that the 2D-QSAR model was able to identify the important molecular fragments and the cross-validation r² (0.794) and standard error (0.127) demonstrated that the 3D-QSAR model was capable of exploring the spatial distribution of important fragments. The obtained results suggested that proposed combination of 2D and 3D QSAR models could be useful in predicting the α-glucosidase inhibiting activity of andrographolide derivatives.

Challenges for blood-brain barrier (BBB) screening

Whilst blood-brain barrier permeability is an important determinant in achieving efficacious central nervous system drug concentrations, it should not be viewed or measured in isolation. Recent studies have highlighted the need for an integrated approach where optimal central nervous system penetration is achieved through the correct balance of permeability, a low potential for active efflux, and the appropriate physicochemical properties that allow for drug partitioning and distribution into brain tissue. Integrating data from permeability studies performed incorporating an assessment of active efflux by P-glycoprotein in combination with drug-free fraction measurements in blood and brain has furthered the understanding of the impact of the blood-brain barrier on central nervous system uptake and the underlying physicochemical properties that contribute to central nervous system drug disposition. This approach moves away from screening and ranking compounds in assays designed to measure or predict central nervous system penetration in the somewhat arbitrary units of brain-blood (or plasma) ratios.

Benchmarking of QSAR Models for Blood-Brain Barrier Permeation

Using the largest available database of 328 blood-brain distribution (logBB) values, a quantitative benchmark was proposed to allow for a consistent comparison of the predictive accuracy of current and future logBB/quantitative structure-activity relationship (-QSAR) models. The usefulness of the benchmark was illustrated by comparing the global and k-nearest neighbors (kNN) multiple-linear regression (MLR) models based on the linear free-energy relationship (LFER) descriptors, and one non-LFER-based MLR model. The leave-one-out (LOO) and leave-group-out Monte Carlo (MC) cross-validation results (q(2) = 0.766, qms = 0.290, and qms(mc) = 0.311) indicated that the LFER-based kNN-MLR model was currently one of the most accurate predictive logBB-QSAR models. The LOO, MC, and kNN-MLR methods have been implemented in the QSAR-BENCH program, which is freely available from www.dmitrykonovalov.org for academic use.

Measuring CAMD Technique Performance. 2. How “Druglike” Are Drugs? Implications of Random Test Set Selection Exemplified Using Druglikeness Classification Models

Research into the advancement of computer-aided molecular design (CAMD) has a tendency to focus on the discipline of algorithm development. Such efforts are often wrought to the detriment of the data set selection and analysis used in said algorithm validation. Here we highlight the potential problems this can cause in the context of druglikeness classification. More rigorous efforts are applied to the selection of decoy (nondruglike) molecules from the ACD. Comparisons are made between model performance using the standard technique of random test set creation with test sets derived from explicit ontological separation by drug class. The dangers of viewing druglike space as sufficiently coherent to permit simple classification are highlighted. In addition the issues inherent in applying unfiltered data and random test set selection to (Q)SAR models utilizing large and supposedly heterogeneous databases are discussed.

In Silico Prediction of Blood−Brain Barrier Permeation Using the Calculated Molecular Cross-Sectional Area as Main Parameter

The cross-sectional area, AD, of a compound oriented in an amphiphilic gradient such as the air-water or lipid-water interface has previously been shown to be crucial for membrane partitioning and permeation, respectively. Here, we developed an algorithm that determines the molecular axis of amphiphilicity and the cross-sectional area, ADcalc, perpendicular to this axis. Starting from the conformational ensemble of each molecule, the three-dimensional conformation selected as the membrane-binding conformation was the one with the smallest cross-sectional area, ADcalcM, and the strongest amphiphilicity. The calculated, ADcalcM, and the measured, AD, cross-sectional areas correlated linearly (n=55, slope, m=1.04, determination coefficient, r2=0.95). The calculated cross-sectional areas, ADcalcM, were then used together with the calculated octanol-water distribution coefficients, log D7.4, of the 55 compounds (with a known ability to permeate the blood-brain barrier) to establish a calibration diagram for the prediction of blood-brain barrier permeation. It yielded a limiting cross-sectional area (ADcalcM=70 A2) and an optimal range of octanol-water distribution coefficients (-1.4<or=log D7.4<7.0). The calibration diagram was validated with an independent set of 43 compounds with the known ability to permeate the blood-brain barrier, yielding a prediction accuracy of 86%. The incorrectly predicted compounds exhibited log D7.4 values comprised between -0.6 and -1.4, suggesting that the limitation for log D7.4 is less rigorous than the limitation for AD. An accuracy of 83% has been obtained for a second validation set of 42 compounds which were previously shown to be difficult to predict. The calculated parameters, ADcalcM and log D7.4, thus allow for a fast and accurate prediction of blood-brain barrier permeation. Analogous calibration diagrams can be established for other membrane barriers.

A data base for partition of volatile organic compounds and drugs from blood/plasma/serum to brain, and an LFER analysis of the data

Literature values of the in vivo distribution (BB) of drugs from blood, plasma, or serum to rat brain have been assembled for 207 compounds (233 data points). We find that data on in vivo distribution from blood, plasma, and serum to rat brain can all be combined. Application of our general linear free energy relationship (LFER) to the 207 compounds yields an equation in log BB, with R2=0.75 and a standard deviation, SD, of 0.33 log units. An equation for a training set predicts the test set of data with a standard deviation of 0.31 log units. We further find that the in vivo data cannot simply be combined with in vitro data on volatile organic and inorganic compounds, because there is a systematic difference between the two sets of data. Use of an indicator variable allows the two sets to be combined, leading to a LFER equation for 302 compounds (328 data points) with R2=0.75 and SD=0.30 log units. A training equation was then used to predict a test set with SD=0.25 log units.

In silico ADME modelling 2: Computational models to predict human serum albumin binding affinity using ant colony systems

Modelling of in vitro human serum albumin (HSA) binding data of 94 diverse drugs and drug-like compounds is performed to develop global predictive models that are applicable to the whole medicinal chemistry space. For this aim, ant colony systems, a stochastic method along with multiple linear regression (MLR), is employed to exhaustively search and select multivariate linear equations, from a pool of 327 molecular descriptors. This methodology helped us to derive optimal quantitative structure-property relationship (QSPR) models based on five and six descriptors with excellent predictive power. The best five-descriptor model is based on Kier and Hall valence connectivity index--Order 5 (path), Auto-correlation descriptor (Broto-Moreau) weighted by atomic masses--Order 4, Auto-correlation descriptor (Broto-Moreau) weighted by atomic polarizabilities--Order 5, AlogP98, SklogS (calculated buffer water solubility) [R=0.8942, Q=0.86790, F=62.24 and SE=0.2626]; the best six-variable model is based on Kier and Hall valence connectivity index of Order 3 (cluster), Auto-correlation descriptor (Broto-Moreau) weighted by atomic masses--Order 4, Auto-correlation descriptor (Broto-Moreau) weighted by atomic polarizabilities--Order 5, Atomic-Level-Based AI topological descriptors--AIdsCH, AlogP98, SklogS (calculated buffer water solubility) [R=0.9128, Q=0.89220, F=64.09 and SE=0.2411]. From the analysis of the physical meaning of the selected descriptors, it is inferred that the binding affinity of small organic compounds to human serum albumin is principally dependent on the following fundamental properties: (1) hydrophobic interactions, (2) solubility, (3) size and (4) shape. Finally, as the models reported herein are based on computed properties, they appear to be a valuable tool in virtual screening, where selection and prioritisation of candidates is required.

Classification Tree Models for the Prediction of Blood−Brain Barrier Passage of Drugs

The use of classification trees for modeling and predicting the passage of molecules through the blood-brain barrier was evaluated. The models were built and evaluated using a data set of 147 molecules extracted from the literature. In the first step, single classification trees were built and evaluated for their predictive abilities. In the second step, attempts were made to improve the predictive abilities using a set of 150 classification trees in a boosting approach. Two boosting algorithms, discrete and real adaptive boosting, were used and compared. High-predictive classification trees were obtained for the data set used, and the models could be improved with boosting. In the context of this research, discrete adaptive boosting gives slightly better results than real adaptive boosting.

In silico prediction of blood brain barrier permeability: an Artificial Neural Network model

This paper has two objectives: first to develop an in silico model for the prediction of blood brain barrier permeability of new chemical entities and second to find the role of active transport specific to the P-glycoprotein (P-gp) substrate probability in blood brain barrier permeability. An Artificial Neural Network (ANN) model has been developed to predict the ratios of the steady-state concentrations of drugs in the brain to those in the blood (logBB) from their molecular structural parameters. Seven descriptors including P-gp substrate probability have been used for model development. The developed model is able to capture a relationship between P-gp and logBB. The predictive ability of the ANN model has also been compared with earlier computational models.