SYNOPSIS
Skin sensitization is also related to the dental field. In addition, the regeneration of skin and mucous membranes of the lips may be consulted from the patient. In silico assessment of skin sensitization is increasingly needed owing to the problems concerning animal welfare, as well as excessive time consumed and cost involved in the development and testing of new chemicals. We could perfectly classify skin sensitizers (positive/negative) using a newly developed K-step Yard sampling (KY) methods (U.S. Patent No. 7725413, 2010). Therefore, the KY methods could be applied to qualitative structure-toxicity relationships (QSTR) study on classifying and predicting samples.
A total of 593 compounds (419 positive sensitizers and 174 negative sensitizers) were used in this study. Parameters were generated from 2-D and 3-D structures of compounds. All of the 1015 parameters generated were reduced by various feature selection methods. KY methods were performed using ADMEWORKS/ModelBuilder software. All 593 compounds were perfectly classified by 3 steps. Discriminant function of each step was a linear dicriminant function, the Iterative Least Squares linear discriminant (TILSQ). KY methods were referred to as a meta-algorithm approach because it requires ordinary data analysis methods to generate discriminant functions.
KY methods were the repetition of removal of gray zone of samples and reclassification of them to attain no gray zone (100% classification) at final step. This methods always attain perfect classification at final step, even though samples are large number, large of structural diversity or highly overlapped on the sample space.
KY methods are promising tool in QSTR technology.
Key words: skin sensitization, qualitative structure-toxicity relationships (QSTR), K-step Yard sampling (KY) methods, animal study