An in silico ensemble method for lead discovery: decision forest

Recent progress in combinatorial chemistry and parallel synthesis has radically changed the approach to drug discovery in the pharmaceutical industry. At present, thousands of compounds can be made in a short period, creating a need for fast and effective in silico methods to select the most promising lead candidates. Decision forest is a novel pattern recognition method, which combines the results of multiple distinct but comparable decision tree models to reach a consensus prediction. In this article, a decision forest model was developed using a structurally diverse training data set containing 232 compounds whose estrogen receptor binding activity was tested at the U.S. Food and Drug Administration (FDA)'s National Center for Toxicological Research (NCTR). The model was subsequently validated using a test data set of 463 compounds selected from the literature, and then applied to a large data set with 57,145 compounds as a screening example. The results show that the decision forest method is a fast, reliable and effective in silico approach, which could be useful in drug discovery.     

Evaluation of a method for high throughput solubility determination using a multi-wavelength UV plate reader

Aqueous solubility is a critical physicochemical property and must be addressed early during drug discovery research. Due to the difficulty in accurately predicting aqueous solubility in silico, high throughput experimental determination of aqueous solubility is in great demand. This study evaluates a method using a multi-wavelength UV plate reader and disposable 96-well UV plates for fast solubility determination. It was demonstrated that this method has the sensitivity and reproducibility to effectively determine solubility as low as 1 micro M. Excellent correlation (R>0.97) was observed between the solubility determined using the UV reader method and the HPLC method over the range of 1-1000 micro M for a diverse set of pharmaceutical compounds. In addition to excellent sensitivity and reproducibility, the UV plate reader method also offers the flexibility of being able to determine thermodynamic solubility in the presence or absence of dimethyl sulfoxide, which is a solvent widely used for combinatorial compounds during high throughput screening.     

Applications of high throughput microsomal stability assay in drug discovery

High throughput in vitro microsomal stability assays are widely used in drug discovery as an indicator for in vivo stability, which affects pharmacokinetics. This is based on in-depth research involving a limited number of model drug-like compounds that are cleared predominantly by cytochrome P450 metabolism. However, drug discovery compounds are often not drug-like, are assessed with high throughput assays, and have many potential uncharacterized in vivo clearance mechanisms. Therefore, it is important to determine the correlation between high throughput in vitro microsomal stability data and abbreviated discovery in vivo pharmacokinetics study data for a set of drug discovery compounds in order to have evidence for how the in vitro assay can be reliably applied by discovery teams for making critical decisions. In this study the relationship between in vitro single time point high throughput microsomal stability and in vivo clearance from abbreviated drug discovery pharmacokinetics studies was examined using 306 real world drug discovery compounds. The results showed that in vitro Phase I microsomal stability t(1/2) is significantly correlated to in vivo clearance with a p-value<0.001. For compounds with low in vitro rat microsomal stability (t(1/2)<15 min), 87% showed high clearance in vivo (CL>25 mL/min/kg). This demonstrates that high throughput microsomal stability data are very effective in identifying compounds with significant clearance liabilities in vivo. For compounds with high in vitro rat microsomal stability (t(1/2)>15 min), no significant differentiation was observed between high and low clearance compounds. This is likely owing to other clearance pathways, in addition to cytochrome P450 metabolism that enhances in vivo clearance. This finding supports the strategy used by medicinal chemists and drug discovery teams of applying the in vitro data to triage compounds for in vivo PK and efficacy studies and guide structural modification to improve metabolic stability. When in vitro and in vivo data are both available for a compound, potential in vivo clearance pathways can be diagnosed to guide further discovery studies.     

Binary classification of aqueous solubility using support vector machines with reduction and recombination feature selection

Aqueous solubility is recognized as a critical parameter in both the early- and late-stage drug discovery. Therefore, in silico modeling of solubility has attracted extensive interests in recent years. Most previous studies have been limited in using relatively small data sets with limited diversity, which in turn limits the predictability of derived models. In this work, we present a support vector machines model for the binary classification of solubility by taking advantage of the largest known public data set that contains over 46?000 compounds with experimental solubility. Our model was optimized in combination with a reduction and recombination feature selection strategy. The best model demonstrated robust performance in both cross-validation and prediction of two independent test sets, indicating it could be a practical tool to select soluble compounds for screening, purchasing, and synthesizing. Moreover, our work may be used for comparative evaluation of solubility classification studies ascribe to the use of completely public resources.     

Noise reduction method for molecular interaction energy: application to in silico drug screening and in silico target protein screening

We developed a new method to improve the accuracy of molecular interaction data using a molecular interaction matrix. This method was applied to enhance the database enrichment of in silico drug screening and in silico target protein screening using a protein-compound affinity matrix calculated by a protein-compound docking software. Our assumption was that the protein-compound binding free energy of a compound could be improved by a linear combination of its docking scores with many different proteins. We proposed two approaches to determine the coefficients of the linear combination. The first approach is based on similarity among the proteins, and the second is a machine-learning approach based on the known active compounds. These methods were applied to in silico screening of the active compounds of several target proteins and in silico target protein screening.     

High-throughput prediction of blood-brain partitioning: a thermodynamic approach

A high-throughput in silico screening tool for potentially CNS active compounds was developed on the basis of the correlation of solvation free energies and blood-brain partitioning (log(cbrain/cblood) = log BB) data available from experimental sources. Utilizing a thermodynamic approach, solvation free energies were calculated by the fast and efficient generalized Born/surface area continuum solvation model, which enabled us to evaluate more than 10 compounds/min. Our training set involved a structurally diverse set of 55 compounds and yielded a function of log BB = 0.035Gsolv + 0.2592 (r = 0.85, standard error 0.37). Calculation of solvation free energies for 8700 CNS active compounds (CIPSLINE database) revealed that Gsolv is higher than -50 kJ/mol for the 96% of these compounds which can be used as suitable criteria for the identification of compounds preferable for CNS penetration.     

Classification of highly unbalanced CYP450 data of drugs using cost sensitive machine learning techniques

In this paper, we study the classifications of unbalanced data sets of drugs. As an example we chose a data set of 2D6 inhibitors of cytochrome P450. The human cytochrome P450 2D6 isoform plays a key role in the metabolism of many drugs in the preclinical drug discovery process. We have collected a data set from annotated public data and calculated physicochemical properties with chemoinformatics methods. On top of this data, we have built classifiers based on machine learning methods. Data sets with different class distributions lead to the effect that conventional machine learning methods are biased toward the larger class. To overcome this problem and to obtain sensitive but also accurate classifiers we combine machine learning and feature selection methods with techniques addressing the problem of unbalanced classification, such as oversampling and threshold moving. We have used our own implementation of a support vector machine algorithm as well as the maximum entropy method. Our feature selection is based on the unsupervised McCabe method. The classification results from our test set are compared structurally with compounds from the training set. We show that the applied algorithms enable the effective high throughput in silico classification of potential drug candidates.     

Analysis of large screening data sets via adaptively grown phylogenetic-like trees

As the use of high-throughput screening systems becomes more routine in the drug discovery process, there is an increasing need for fast and reliable analysis of the massive amounts of the resulting data. At the forefront of the methods used is data reduction, often assisted by cluster analysis. Activity thresholds reduce the data set under investigation to manageable sizes while clustering enables the detection of natural groups in that reduced subset, thereby revealing families of compounds that exhibit increased activity toward a specific biological target. The above process, designed to handle primarily data sets of sizes much smaller than the ones currently produced by high-throughput screening systems, has become one of the main bottlenecks of the modern drug discovery process. In addition to being fragmented and heavily dependent on human experts, it also ignores all screening information related to compounds with activity less than the threshold chosen and thus, in the best case, can only hope to discover a subset of the knowledge available in the screening data sets. To address the deficiencies of the current screening data analysis process the authors have developed a new method that analyzes thoroughly large screening data sets. In this report we describe in detail this new approach and present its main differences with the methods currently in use. Further, we analyze a well-known, publicly available data set using the proposed method. Our experimental results show that the proposed method can improve significantly both the ease of extraction and amount of knowledge discovered from screening data sets.     

Collection and preparation of molecular databases for virtual screening

Drug discovery and development research is undergoing a paradigm shift from a linear and sequential nature of the various steps involved in the drug discovery process of the past to the more parallel approach of the present, due to a lack of sufficient correlation between activities estimated by in vitro and in vivo assays. This is attributed to the non-drug-likeness of the lead molecules, which has often been detected at advanced drug development stages. Thus a striking aspect of this paradigm shift has been early/parallel in silico prioritization of drug-like molecular databases (also database pre-processing), in addition to prioritizing compounds with high affinity and selectivity for a protein target. In view of this, a drug-like database useful for virtual screening has been created by prioritizing molecules from 36 catalog suppliers, using our recently derived binary QSAR based drug-likeness model as a filter. The performance of this model was assessed by a comparative evaluation with respect to commonly used filters implemented by the ZINC database. Since the model was derived considering all the limitations that have plagued the existing rules and models, it performs better than the existing filters and thus the molecules prioritized by this filter represent a better subset of drug-like compounds. The application of this model on exhaustive subsets of 4,972,123 molecules, many of which have passed the ZINC database filters for drug-likeness, led to a further prioritization of 2,920,551 drug-like molecules. This database may have a great potential for in silico virtual screening for discovering molecules, which may survive the later stages of the drug development research.     

Prediction of ion channel activity using binary kernel discrimination

Voltage-gated ion channels are a diverse family of pharmaceutically important membrane proteins for which limited 3D information is available. A number of virtual screening tools have been used to assist with the discovery of new leads and with the analysis of screening results. One such tool, and the subject of this paper, is binary kernel discrimination (BKD), a machine-learning approach that has recently been applied to applications in chemoinformatics. It uses a training set of compounds, for which both structural and qualitative activity data are known, to produce a model that can then be used to rank another set of compounds in order of likely activity. Here, we report the use of BKD to build models for the prediction of five different ion channel targets using two types of activity data. The results obtained suggest that the approach provides an effective way of prioritizing compounds for acquisition and testing.     

Data mining the NCI60 to predict generalized cytotoxicity

Elimination of cytotoxic compounds in the early and later stages of drug discovery can help reduce the costs of research and development. Through the application of principal components analysis (PCA), we were able to data mine and prove that approximately 89% of the total log GI 50 variance is due to the nonspecific cytotoxic nature of substances. Furthermore, PCA led to the identification of groups of structurally unrelated substances showing very specific toxicity profiles, such as a set of 45 substances toxic only to the Leukemia_SR cancer cell line. In an effort to predict nonspecific cytotoxicity on the basis of the mean log GI 50, we created a decision tree using MACCS keys that can correctly classify over 83% of the substances as cytotoxic/noncytotoxic in silico, on the basis of the cutoff of mean log GI 50 = -5.0. Finally, we have established a linear model using least-squares in which nine of the 59 available NCI60 cancer cell lines can be used to predict the mean log GI 50. The model has R (2) = 0.99 and a root-mean-square deviation between the observed and calculated mean log GI 50 (RMSE) = 0.09. Our predictive models can be applied to flag generally cytotoxic molecules in virtual and real chemical libraries, thus saving time and effort.     

In Silico Identification of Tripeptides as Lead Compounds for the Design of KOR Ligands

The kappa opioid receptor (KOR) represents an attractive target for the development of drugs as potential antidepressants, anxiolytics and analgesics. A robust computational approach may guarantee a reduction in costs in the initial stages of drug discovery, novelty and accurate results. In this work, a virtual screening workflow of a library consisting of ~6 million molecules was set up, with the aim to find potential lead compounds that could manifest activity on the KOR. This in silico study provides a significant contribution in the identification of compounds capable of interacting with a specific molecular target. The main computational techniques adopted in this experimental work include: (i) virtual screening; (ii) drug design and leads optimization; (iii) molecular dynamics. The best hits are tripeptides prepared via solution phase peptide synthesis. These were tested in vivo, revealing a good antinociceptive effect after subcutaneous administration. However, further work is due to delineate their full pharmacological profile, in order to verify the features predicted by the in silico outcomes.     

Recent Advances in In Silico Target Fishing

In silico target fishing, whose aim is to identify possible protein targets for a query molecule, is an emerging approach used in drug discovery due its wide variety of applications. This strategy allows the clarification of mechanism of action and biological activities of compounds whose target is still unknown. Moreover, target fishing can be employed for the identification of off targets of drug candidates, thus recognizing and preventing their possible adverse effects. For these reasons, target fishing has increasingly become a key approach for polypharmacology, drug repurposing, and the identification of new drug targets. While experimental target fishing can be lengthy and difficult to implement, due to the plethora of interactions that may occur for a single small-molecule with different protein targets, an in silico approach can be quicker, less expensive, more efficient for specific protein structures, and thus easier to employ. Moreover, the possibility to use it in combination with docking and virtual screening studies, as well as the increasing number of web-based tools that have been recently developed, make target fishing a more appealing method for drug discovery. It is especially worth underlining the increasing implementation of machine learning in this field, both as a main target fishing approach and as a further development of already applied strategies. This review reports on the main in silico target fishing strategies, belonging to both ligand-based and receptor-based approaches, developed and applied in the last years, with a particular attention to the different web tools freely accessible by the scientific community for performing target fishing studies.     

'LC-electrolyte effects' improve the bioanalytical performance of liquid chromatography/tandem mass spectrometric assays in supporting pharmacokinetic study for drug discovery

The development of rapid and sensitive bioanalytical methods in a short time frame with acceptable levels of precision and accuracy is imperative for successful drug discovery. We previously reported that the use of a mobile phase containing an extremely low concentration of ammonium formate or formic acid increased analyte electrospray ionization (ESI) response and controlled against matrix effects. We designated these favorable effects 'LC-electrolyte effects'. In order to support rapid pharmacokinetic (PK) studies for drug discovery, we applied LC-electrolyte effects to the development of generic procedures that can be used to quickly generate reliable PK data for compound candidates. We herein demonstrate our approach using four model tested compounds (Compd-A, -B, -C, and -D). The analytical methods involve generic protein precipitation for sample clean-up, followed by application of fast liquid chromatographic (LC) gradients and the subsequent use of electrospray ionization tandem mass spectrometry (ESI-MS/MS) for individual measurement of the tested compounds in 20-microL plasma samples. Good linearity over the concentration range of 1.6 or 8-25000 ng/mL (r(2) > 0.99), precision (RSD, 0.45-13.1%), and accuracy (91-112%) were achieved through the use of a low dose of formic acid (0.4 mM or 0.015 per thousand) in the methanol/water-based LC mobile phase. The analytical method was quite sensitive, providing a lower limit of quantification of 1.6 pg on-column except for Compd-C (8 pg), and showed negligible ion suppression caused by matrix components. Finally, the assay suitability was demonstrated in simulated discovery PK studies of the tested compounds with i.v./p.o. dosing of rats. This new assay approach has been adopted with good results in our laboratory for many recent discovery PK studies.     

Integrated decision support for assessing chemical liabilities

Chemical liabilities, such as adverse effects and toxicity, have a major impact on today's drug discovery process. In silico prediction of chemical liabilities is an important approach which can reduce costs and animal testing by complementing or replacing in vitro and in vivo liability models. There is a lack of integrated, extensible decision support systems for chemical liability assessment which run quickly and have easily interpretable results. Here we present a method which integrates similarity searches, structural alerts, and QSAR models which all are available from the Bioclipse workbench. Emphasis has been placed on interpretation of results, and substructures which are important for predictions are highlighted in the original chemical structures. This allows for interactively changing chemical structures with instant visual feedback and can be used for hypothesis testing of single chemical structures as well as compound collections. The system has a clear separation between methods and data, and the extensible architecture enables straightforward extension via addition of more plugins (such as new data sets and computational models). We demonstrate our method on three important safety end points: mutagenicity, carcinogenicity, and aryl hydrocarbon receptor (AhR) activation. Bioclipse and the decision support implementation are free, open source, and available from http://www.bioclipse.net/decision-support .     

Optimizing Dvl PDZ domain inhibitor by exploring chemical space

Because of advances in the high-throughput screening technology, identification of a hit that can bind to a target protein has become a relatively easy task; however, in the process of drug discovery, the following hit-to-lead and lead optimization still remain challenging. In a typical hit-to-lead and lead optimization process, the analogues of the most promising hits are synthesized for the development of structure–activity relationship (SAR) analysis, and in turn, in the effort of optimization of lead compounds, such analysis provides guidance for the further synthesis. The synthesis processes are usually long and labor-intensive. In silico searching has becoming an alternative approach to explore SAR especially with millions of compounds ready to be screened and most of them can be easily obtained. Here, we report our discovery of 15 new Dishevelled PDZ domain inhibitors by using such an approach. In our studies, we first developed a pharmacophore model based on NSC668036, an inhibitor previously identified in our laboratory; based on the model, we then screened the ChemDiv database by using an algorithm that combines similarity search and docking procedures; finally, we selected potent inhibitors based on docking analysis and examined them by using NMR spectroscopy. NMR experiments showed that all the 15 compounds we chose bound to the PDZ domain tighter than NSC668036.