Related presentations, posters, and scientific talks from this event have been posted here for your reference. Please click the associated link to download.
| Title | Author | Link |
|---|---|---|
| Classification of Drugs by CNS Activity Based on QSAR Models of the Rate and Extent of Brain Delivery | Kiril Lanevskij; Pranas Japertas; Remigijus Didziapetris; Alanas Petrauskas | Download the poster |
| GALAS Modeling Methodology Applications in the Prediction of Drug Safety Related Properties | Andrius Sazonovas; Remigijus Didziapetris; Justas Dapkunas; Liutauras Juska; and Pranas Japertas | Download the poster |
| Probabilistic Model of Regioselectivity of Metabolism in Human Liver Microsomes | Justas Dapkunas; Andrius Sazonovas; Pranas Japertas | Download the poster |
Booth # 621
Title: Probabilistic model of regioselectivity of metabolism in human liver microsomes
Date: Wednesday, March 24, 2010
Time: General Poster Session, 7:00–9:00 PM
Abstract #: 437
Authors: Justas Dapkunas, Andrius Sazonovas, Liutauras Juska, Pranas Japertas
Abstract: Here we present a model for in silico prediction of the most probable sites of human liver microsomal metabolism in a molecule. The developed models calculate the probabilities of being a target of human cytochrome P450 enzymes (CYP3A4, CYP2D6, CYP2C9, CYP2C19, CYP1A2) for any atom in a molecule and allow forecasting of the most probable phase I metabolites. The novel GALAS (Global, Adjusted Locally According to Similarity) modeling methodology was used for development of probabilistic models. It provides a possibility to estimate the reliability of prediction. Moreover, the Applicability Domain of the models can be easily expanded to cover compound classes of user interest by incorporating 'in house' databases containing experimental metabolism data. Experimental data for >600 compounds with >6000 different carbon atoms were used for modeling. Four baseline models were developed for four types of atoms (aromatic carbon, aliphatic carbon, carbon near nitrogen, carbon near oxygen). GALAS modeling methodology finds most similar metabolism sites in the training set, according to which corrections to the baseline predictions are made and final prediction quality is estimated in the form of Reliability Index (RI). The numbers of mispredictions and inconclusive results reduce significantly when only results of high quality (RI>0.5) are taken into account, demonstrating that RI reflects accuracy of prediction. The regioselectivity models are shown to be trainable using experimental data for compounds not present in the training set.
Title: Classification of drugs according to presence or absence of CNS activity based on mechanistic QSAR models of the rate and extent of brain delivery
Date: Wednesday, March 24, 2010
Time: General Poster Session, 7:00–9:00 PM
Abstract #: 494
Authors: Kiril Lanevskij, Pranas Japertas, Remigijus Didziapetris
Abstract: This study presents a simple classification scheme for in silico evaluation whether brain penetration of novel compounds is sufficient to exhibit central action. Classification is performed taking into account both kinetic and thermodynamic characteristics of drug transport across blood-brain barrier. The calculation procedure for the brain penetration rate expressed by permeability-surface area product (log PS) was described previously (Lanevskij K, Japertas P, Didziapetris R, Petrauskas A. Ionization-specific prediction of blood-brain permeability. J Pharm Sci. 2009 Jan;98(1):122-34). The extent of brain penetration was represented by experimentally determined steady-state brain/blood distribution ratios (log BB) for about 500 compounds collected from literature. These data were split into two terms corresponding to drug binding to plasma proteins and brain constituents - the two major processes that influence partitioning between brain and plasma under the assumption of passive diffusion-driven transport. Brain tissue binding affinity of drugs was then described by a nonlinear model in terms of key physicochemical determinants - octanol/water log P and pKa. Prediction of CNS activity was performed on the basis of calculated log BB and supplementary parameter brain/plasma equilibration rate defined as log PS corrected for unbound fraction in brain. It was shown that a simple combination of the respective models allows correctly classifying more than 90% of drugs in the literature data set comprised of about 1600 diverse molecules with experimentally assigned CNS activity categories (CNS+/CNS-). Moreover, as demonstrated by several examples, the proposed classification scheme provides an insight on the onset and duration of action of central drugs.
Title: GALAS modeling methodology applications in the prediction of the drug safety related properties
Date: Wednesday, March 24, 2010
Time: General Poster Session, 7:00–9:00 PM
Abstract #: 432
Authors: Andrius Sazonovas, Remigijus Didziapetris, Justas Dapkunas, Liutauras Juska, Pranas Japertas
Abstract: Early computational evaluation of drug candidate properties related to its pharmaceutical safety is becoming increasingly important in the drug discovery process. Yet the effective use of any available third-party predictive algorithms for these properties in the pharmaceutical industry is severely hindered by a number of problems. E.g. the training set rarely covers the specific part of the chemical space occupied by the compounds that a certain company is working with or a specific experimental protocol is used to measure the corresponding properties or activities 'in house'. Therefore the need arises for a method that would allow any company to tailor a third-party predictive algorithm to its specific needs using proprietary 'in house' data. Here we present a novel GALAS (Global, Adjusted Locally According to Similarity) modeling methodology that provides a possibility for a researcher to expand the Applicability Domain of the resulting models with the help of a custom database of experimental values for the property of interest. A Reliability Index (RI) is also calculated as a measure of the quality of the particular prediction. The use of the method is illustrated with examples of its application in predicting CYP3A4 and hERG inhibition which figure among the major factors attributing to the rising attrition rate, being responsible for the various unwanted drug-drug interactions and cardiotoxicity respectively. It is shown that a relatively small amount (5 to 10) of similar compounds has to be added to substantially improve the prediction for a group of problematic compounds that is not represented in the original training set. Similarly the models are shown to be able to utilize 'in house' data obtained using different protocol compared the experimental training set data. Most importantly all of the above benefits are obtained without time consuming statistical retraining of the initial GALAS models.
