scispace - formally typeset
Search or ask a question

Showing papers by "Nina Jeliazkova published in 2009"


Journal ArticleDOI
TL;DR: The factors that modulate carcinogenic potency are examined: this is an additional piece of information that can have a role in fine‐tuning a risk assessment.
Abstract: The structure alerts (SA) for carcinogenicity/mutagenicity are a repository of the science on chemical biological interactions; in addition, they have a crucial role in practical applications for risk assessment. In predictive toxicology, it is crucial that knowledge of SAs is accompanied by knowledge of the structural motifs that modulate their effects. Recently, we have compiled an updated list of SAs implemented in the expert system Toxtree 1.50 (open source, freely available). These SAs are aimed at discriminating between active and inactive chemicals, and include only modulating factors with a high probability of eliminating completely the effect of the SA. Here we have examined the factors that modulate carcinogenic potency: this is an additional piece of information that can have a role in fine-tuning a risk assessment. The case study selected is the carcinogenic potential of the aromatic amines in rats and mice. As the carcinogenic potency of these compounds is different in mice and rats (correlation coefficient = 0.546), there are both agreements and differences in the pattern of these motifs. Differences are observed mainly for the motifs that decrease the carcinogenic potency of aromatic amines. In mice, substitutions ortho and meta to the amino group tend to decrease the potency, as well as −NO2 in any position. In rats, these motifs affect the potency to a more limited extent. On the other hand, increasing effects are quite similar in the two animals and are exerted mainly by additional rings, tricyclic systems, five-numbered rings, and N-heteroaromatic systems. Environ. Mol. Mutagen. 2009. © 2009 Wiley-Liss, Inc.

12 citations


Journal ArticleDOI
TL;DR: In this paper, the authors compare the results from Bayesian classification based on non-parametric probability density estimates of the data to the results obtained from other classification methods, making use of a small benchmark dataset, a larger dataset from Corine land cover project for Bulgaria and analyzing different features and feature selection methods.

2 citations