Loading...

QSAR Modeling of Inhibition Behavior of Diabetes Type II Inhibitors

Bigdeli, Arafeh | 2011

573 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 42198 (03)
  4. University: Sharif University of Technology
  5. Department: chemistry
  6. Advisor(s): Jalali Heravi, Mehdi
  7. Abstract:
  8. QSAR studies play an important role in drug design and lead optimization by developing a mathematical relationship between the chemical structures of compounds and their biological activities.By using such an approach one could predict the activities of newly designed compounds before a decision is made whether these compounds should be really synthesized or tested. The main aim of the present work was to develop a QSAR model for predicting the inhibitory activity of acetyl CoA carboxylase (ACC)derivatives as anti-diabetic inhibitors. Since variable selection is a critical step in every QSAR study, four different algorithms, based on Monte Carlo cross-validation techniques, were investigated. Comparing the results of these techniques (including stepwise MLR-ANN, SPA-ANN, GA-MLR-ANN and GA-ANN), dedicates that the GA-ANN (better called BRGNN) model is able to select the best variables for predicting the inhibition action of the ACC inhibitors. In addition to non-linear ANN modeling, the preceding procedure was repeated for linear MLR modeling. Superiority of the non-linear over the linear models revealed that the inhibitory behavior has non-linear characteristics. A sensitivity analysis was done to characterize the relative importance of the descriptors. By representing the most important descriptor in ANN modeling and ranking the present variables by importance, we have alleviated the black box nature of the neural network methodology to some extent. The most important variable was found to be RDF050m, which is proposed to evaluate the three dimensional distribution of atomic masses. Therefore, together with the other selected descriptors, it could be suggested that the inhibition mechanism of ACC compounds relies on the mass geometrical 3D-distribution of the atoms in a non-linear manner.
  9. Keywords:
  10. Genetic Algorithm ; Quantitative Structure-Activity Relationship (QSAR)Model ; Anti-Diabetic Inhibitors ; Acetyle CoA Carboxylase ; Successive Projections Algorithm

 Digital Object List

 Bookmark

No TOC