Please use this identifier to cite or link to this item:
|Title:||Development and validation of a robust QSAR model for prediction of carcinogenicity of drugs|
|Abstract:||Carcinogenicity is one of the toxicological endpoints causing the highest concern. Also, the standard bioassays in rodents used to assess the carcinogenic potential of chemicals and drugs are extremely long, costly and require the sacrifice of large numbers of animals. For these reasons, we have attempted development of a global quantitative structure–activity relationship (QSAR) model using a data set of 1464 compounds (the Galvez data set available from http://www.uv.es/~galvez/tablevi.pdf), including many marketed drugs for their carcinogenesis potential. Though experimental toxicity testing using animal models is unavoidable for new drug candidates at an advanced stage of drug development, yet the developed global QSAR model can in silico predict the carcinogenicity of new drug compounds to provide a tool for initial screening of new drug candidate molecules with reduced number of animal testing, money and time. Considering large number of data points with diverse structural features used for model development (ntraining = 732) and model validation (ntest = 732), the model developed in this study has an encouraging statistical quality (leave-one-out Q2 = 0.731, R2pred = 0.716). Our developed model suggests that higher lipophilicity values and conjugated ring systems, thioketo and nitro groups contribute positively towards drug carcinogenicity. On the contrary, tertiary and secondary nitrogens, phenolic, enolic and carboxylic OH fragments and presence of three-membered rings reduce the carcinogenicity. Branching, size and shape are found to be crucial factors for drug-induced carcinogenicity. One may consider all these points to reduce carcinogenic potential of the molecules.|
|ISSN:||0975-0959 (Online); 0301-1208 (Print)|
|Appears in Collections:||IJBB Vol.48(2) [April 2011]|
Items in NOPR are protected by copyright, with all rights reserved, unless otherwise indicated.