|
Knowledge Fair Table B - Drug Discovery Application
Posted by
Roman Affentranger
at
Sep 13, 2010 08:55 PM
|
Permalink
An example of a predictive toxicology application in drug discovery is given using the data on antimalarial compounds made available at the ChEMBL Neglected Tropical Disease (NTD) archive (http://www.ebi.ac.uk/chemblntd/). The data was imported and made available via a specially installed OpenTox dataset service, to emulate a drug discovery exercise working in a setup separate from public dataset services. OpenTox model services are used for predicting toxicities. For this workshop activity, we combined all three data sets from the ChEMBL NTD archive in a newly created OpenTox data service: the Tres Cantos Antimalarial TCAMS dataset (>13'000 antimalarials found in a 2-million compound screening library of GlaxoSmithKline), the Novartis-GNF dataset (>5'000 antimalarials) and the St. Jude Children's Research Hospital dataset (>1'100 antimalarials, 172 of which were studied in great detail). More information on these datasets can be found at http://www.ebi.ac.uk/chemblntd/. Workshop Activity:
A) Prioritizing the AntimalarialsStep 1The antimalarial compounds are prioritized based on a very conservative model for predicting oral toxicity. More specifically, Cramer rules (a model for predicting oral toxicity) are applied to all datasets. Cramer rules is a very conservative model, that is, if it predicts a chemical to be save (class I), there is high probability it will be safe indeed. Food ingredients, for example, are classified as Class I. Step 2Experimentally determined cytotoxicities against human cells of the compounds predicted to be safe (Cramer class I) are further examined, and their mutagenicities predicted via Toxtree Benigni Bossa rules. Step 3A new dataset is created with compounds that have no mutagenicity alerts, low human cytotoxicity, but high anti-malarial activity. Potential sites of cytochrome P450 metabolism are predicted for for this dataset. Download detailed instructions for this activity here...
B) Additional exercise:Extract subsets of data to be used to create a model via the OpenTox demo application ToxCreate. Download detailed instructions for this activity here... Document Actions |