Development of CYP3A4 inhibition models: comparisons of machine-learning techniques and molecular descriptors

Computational models of cytochrome P450 3A4 inhibition were developed based on high-throughput screening data for 4470 proprietary compounds. Multiple models differentiating inhibitors (IC(50) <3 microM) and noninhibitors were generated using various machine-learning algorithms (recursive partitioning [RP], Bayesian classifier, logistic regression, k-nearest-neighbor, and support vector machine [SVM]) with structural fingerprints and topological indices. Nineteen models were evaluated by internal 10-fold cross-validation and also by an independent test set. Three most predictive models, Barnard Chemical Information (BCI)-fingerprint/SVM, MDL-keyset/SVM, and topological indices/RP, correctly classified 249, 248, and 236 compounds of 291 noninhibitors and 135, 137, and 147 compounds of 179 inhibitors in the validation set. Their overall accuracies were 82%, 82%, and 81%, respectively. Investigating applicability of the BCI/SVM model found a strong correlation between the predictive performance and the structural similarity to the training set. Using Tanimoto similarity index as a confidence measurement for the predictions, the limitation of the extrapolation was 0.7 in the case of the BCI/SVM model. Taking consensus of the 3 best models yielded a further improvement in predictive capability, kappa = 0.65 and accuracy = 83%. The consensus model could also be tuned to minimize either false positives or false negatives depending on the emphasis of the screening.     

"Plate cherry picking": a novel semi-sequential screening paradigm for cheaper, faster, information-rich compound selection

This work describes a novel semi-sequential technique for in silico enhancement of high-throughput screening (HTS) experiments now employed at Novartis. It is used in situations in which the size of the screen is limited by the readout (e.g., high-content screens) or the amount of reagents or tools (proteins or cells) available. By performing computational chemical diversity selection on a per plate basis (instead of a per compound basis), 25% of the 1,000,000-compound screening was optimized for general initial HTS. Statistical models are then generated from target-specific primary results (percentage inhibition data) to drive the cherry picking and testing from the entire collection. Using retrospective analysis of 11 HTS campaigns, the authors show that this method would have captured on average two thirds of the active compounds (IC(50) < 10 microM) and three fourths of the active Murcko scaffolds while decreasing screening expenditure by nearly 75%. This result is true for a wide variety of targets, including G-protein-coupled receptors, chemokine receptors, kinases, metalloproteinases, pathway screens, and protein-protein interactions. Unlike time-consuming "classic" sequential approaches that require multiple iterations of cherry picking, testing, and building statistical models, here individual compounds are cherry picked just once, based directly on primary screening data. Strikingly, the authors demonstrate that models built from primary data are as robust as models built from IC(50) data. This is true for all HTS campaigns analyzed, which represent a wide variety of target classes and assay types.     

A 3D-QSAR model based screen for dihydropyridine-like compound library to identify inhibitors of amyloid beta (Aβ) production

Abnormal accumulation of amyloid beta peptide (Aβ) is one of the hallmarks of Alzheimer's disease progression. Practical limitations such as cost , poor hit rates and a lack of well characterized targets are a major bottle neck in the in vitro screening of a large number of chemical libraries and profiling them to identify Aβ inhibitors. We used a limited set of 1,4 dihydropyridine (DHP)-like compounds from our model set (MS) of 24 compounds which inhibit Aβ as a training set and built 3D-QSAR (Three-dimensional Quantitative Structure-Activity Relationship) models using the Phase program (Schr?dinger, USA). We developed a 3D-QSAR model that showed the best prediction for Aβ inhibition in the test set of compounds and used this model to screen a 1,043 DHP-like small library set of (LS) compounds. We found that our model can effectively predict potent hits at a very high rate and result in significant cost savings when screening larger libraries. We describe here our in silico model building strategy, model selection parameters and the chemical features that are useful for successful screening of DHP and DHP-like chemical libraries for Aβ inhibitors.     

Predicting novel substrates for enzymes with minimal experimental effort with active learning

Enzymatic substrate promiscuity is more ubiquitous than previously thought, with significant consequences for understanding metabolism and its application to biocatalysis. This realization has given rise to the need for efficient characterization of enzyme promiscuity. Enzyme promiscuity is currently characterized with a limited number of human-selected compounds that may not be representative of the enzyme's versatility. While testing large numbers of compounds may be impractical, computational approaches can exploit existing data to determine the most informative substrates to test next, thereby more thoroughly exploring an enzyme's versatility. To demonstrate this, we used existing studies and tested compounds for four different enzymes, developed support vector machine (SVM) models using these datasets, and selected additional compounds for experiments using an active learning approach. SVMs trained on a chemically diverse set of compounds were discovered to achieve maximum accuracies of ~80% using ~33% fewer compounds than datasets based on all compounds tested in existing studies. Active learning-selected compounds for testing resolved apparent conflicts in the existing training data, while adding diversity to the dataset. The application of these algorithms to wide arrays of metabolic enzymes would result in a library of SVMs that can predict high-probability promiscuous enzymatic reactions and could prove a valuable resource for the design of novel metabolic pathways.     

In silico design of potent EGFR kinase inhibitors using combinatorial libraries

This paper is an attempt to design 4-anilinoquinazoline compounds having promising anticancer activities against epidermal growth factor (EGFR) kinase inhibition, using virtual combinatorial library approach. Partial least squares method has been applied for the development of a quantitative structure–activity relationship (QSAR) model based on training and test set approaches. The partial least squares model showed some interesting results in terms of internal and external predictability against EGFR kinase inhibition for such type of anilinoquinazoline derivatives. In virtual screening study, out of 4860 compounds in chemical library, 158 compounds were screened and finally, 10 compounds were selected as promising EGFR kinase inhibitors based on their predicted activities from the QSAR model. These derivatives were subjected to molecular docking study to investigate the mode of binding with the EGFR kinase, and the two compounds (ID 3639 and 3399) showing similar type of docking score and binding patterns with that of the existing drug molecules like erlotinib were finally reported.     

Two simple and generic antibody-independent kinase assays: comparison of a bioluminescent and a microfluidic assay format

In this study, the authors have compared the performance of 2 high-throughput screening assays for a serin/threonine kinase: a microplate-based, bioluminescent assay that uses the luciferin/luciferase system to monitor ATP consumption, and a microfluidic assay that measures the change in mobility in an electric field of a fluorescently labeled peptide upon phosphorylation. Both assays are homogeneous, nonradioactive, antibody independent and could be miniaturized to a reaction volume of 4 microl. The robustness of both formats was demonstrated by Z' values > 0.8. Screening of a small library (2133 compounds) showed that the results obtained with both technologies correlate very well. Although the threshold for hits was set to a comparably low value-22.2% and 13.7% inhibition for the ATP consumption and microfluidic assay, respectively, corresponding to mean plus 3 standard deviations-the overlap of active compounds identified with the 2 assay formats was greater than 94%. Thus, both assays allow the identification of even low potency inhibitors with a high level of confidence.     

High-throughput screening of a 100,000-compound library for inhibitors of influenza A virus (H3N2)

Using a highly reproducible and robust cell-based high-throughput screening (HTS) assay, the authors screened a 100,000-compound library at 14- and 114-microM compound concentration against influenza strain A/Udorn/72 (H3N2). The "hit" rates (>50% inhibition of the viral cytopathic effect) from the 14- and 114-microM screens were 0.022% and 0.38%, respectively. The hits were evaluated for their antiviral activity, cell toxicity, and selectivity in dose-response experiments. The screen at the lower concentration yielded 3 compounds, which displayed moderate activity (SI(50) = 10-49). Intriguingly, the screen at the higher concentration revealed several additional hits. Two of these hits were highly active with an SI(50) > 50. Time of addition experiments revealed 1 compound that inhibited early and 4 other compounds that inhibited late in the virus life cycle, suggesting they affect entry and replication, respectively. The active compounds represent several different classes of molecules such as carboxanilides, 1-benzoyl-3-arylthioureas, sulfonamides, and benzothiazinones, which have not been previously identified as having antiviral/anti-influenza activity.     

Screening for dihydrofolate reductase inhibitors using MOLPRINT 2D, a fast fragment-based method employing the naïve Bayesian classifier: limitations of the descriptor and the importance of balanced chemistry in training and test sets

A fragment-based similarity searching method, MOLPRINT 2D, was employed for virtual screening of Escherichia coli dihydrofolate reductase inhibitors. Using the original training set of 50,000 compounds, only marginal enrichment factors (between 1 and 3) could be achieved on the test library. The active structures contained in the training and test libraries represented different types of "chemistry", that is, different substructural features associated with activity. Training and test sets were pooled in a 2nd step and randomly split into training and test of equal size, with the objective of smoothing out the different chemical characteristics of both libraries. In a 10-fold cross-validation study on the new training and test sets, typically 10-fold enrichment could be found in the first 96 positions, 4-fold enrichment in the first 384 positions, and 3-fold enrichment in the first 1536 positions, corresponding to 6, 10, and 28 hits, respectively (out of a total of 307; activity defined as average residual activity of less than 80%). The conclusions are 2-fold. On one hand, the exact fragment-matching similarity searching method employed here is not capable of finding completely novel hit structures. On the other hand, this study emphasizes the requirement for a comparable distribution of chemical features of the training and test sets. MOLPRINT 2D is freely downloadable from http://www.cheminformatics.org.     

IKKβ inhibitors identification part I: Homology model assisted structure based virtual screening

Control of NF-κB release through the inhibition of IKKβ has been identified as a potential target for the treatment of inflammatory and autoimmune diseases. We have employed structure based virtual screening scheme to identify lead like molecule from ChemDiv database. Homology models of IKKβ enzyme were developed based on the crystal structures of four kinases. The efficiency of the homology model has been validated at different levels. Docking of known inhibitors library revealed the possible binding mode of inhibitors. Besides, the docking sequence analyses results indicate the responsibility of Glu172 in selectivity. Structure based virtual screening of ChemDiv database has yielded 277 hits. Top scoring 75 compounds were selected and purchased for the IKKβ enzyme inhibition test. From the combined approach of virtual screening followed by biological screening, we have identified six novel compounds that can work against IKKβ, in which 1 compound had highest inhibition rate 82.09% at 10 μM and IC₅₀ 1.76 μM and 5 compounds had 25.35–48.80% inhibition.     

High-throughput screening by mass spectrometry: comparison with the scintillation proximity assay with a focused-file screen of AKT1/PKB alpha

Mass spectrometry is an emerging format for label-free high-throughput screening. The main limitation of mass spectrometry is throughput, due to the requirement to purify samples prior to ionization. Here the authors compare an automated high-throughput mass spectrometry (HTMS) system (RapidFire) with the scintillation proximity assay (SPA). The cancer therapy target AKT1/PKBalpha was screened against a focused library of kinase inhibitors and IC50 values determined for all compounds that exhibit > 50% inhibition. A selection of additional compounds that exhibited 相似文献   

Statistical molecular design of a focused salicylidene acylhydrazide library and multivariate QSAR of inhibition of type III secretion in the Gram-negative bacterium Yersinia

A combined application of statistical molecular design (SMD), quantitative structure–activity relationship (QSAR) modeling and prediction of new active compounds was effectively used to develop salicylidene acylhydrazides as inhibitors of type III secretion (T3S) in the Gram-negative pathogen Yersinia pseudotuberculosis. SMD and subsequent synthesis furnished 50 salicylidene acylhydrazides in high purity. Based on data from biological evaluation in T3S linked assays 18 compounds were classified as active and 25 compounds showed a dose-dependent inhibition. The 25 compounds were used to compute two multivariate QSAR models and two multivariate discriminant analysis models were computed from both active and inactive compounds. Three of the models were used to predict 4416 virtual compounds in consensus and eight new compounds were selected as an external test set. Synthesis and biological evaluation of the test set in Y. pseudotuberculosis and the intracellular pathogen Chlamydia trachomatis validated the models. Y. pseudotuberculosis and C. trachomatis cell-based infection models showed that compounds identified in this study are selective and non-toxic inhibitors of T3S dependent virulence.