Mutagenicity: QSAR - quasi-QSAR - nano-QSAR.

Mutagenic potential of biphenyl-4-amines and multi-walled carbon nanotubes (MWCNTs) have been modeled by optimal descriptors. The optimal descriptors are calculated with the Monte Carlo method by means of the CORAL software (http://www.insilico.eu/coral). The optimal descriptor is a translator of eclectic data into prediction of various endpoints in general and into the prediction of the mutagenic potential (TA100) in particular. So-called, quasi-SMILES are suggested as representation of various circumstances which can influence the endpoint. The correlation weights of various circumstances are the basis of the approach. The statistical characteristics of models for the mutagenic potential (decimal logarithm of TA100) for the external invisible validation sets are the following (i) in the case of biphenyl-4- amines: n=7-11; r(2)=0.649±0.046; s=0.211±0.029; and (ii) in the case of MWCNTs: n=6, r(2)=0.804±0.107; s=0.048±0.01.