Loading...

The Use of Enhanced Chemometric Methods in QSAR and Pattern Recognition Studies

Asadollahi Babol, Mohammad | 2010

742 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 40223 (03)
  4. University: Sharif University of Technology
  5. Department: Chemistry
  6. Advisor(s): Jalali Heravi, Mehdi
  7. Abstract:
  8. Chemometrics is an interdisciplinary subject which has many applications among science and industrial process. Environmental chemist, Food chemist, biologist and so on depend on good analytical chemistry measurement and so need chemometrics to interpret their data. In this project, we have developed different chemometrics and machine learning methods for considering the quantitative structure-activity relationship between different drug-like molecules. Also some chemometrics techniques were applied for the pattern recognition of NIR data on human plasma for discriminating between healthy and infected HIV-1 patients. At first, Quantitative structure-activity relationship (QSAR) models for inhibition action of some 1-phenylbenzimidazoles on platelet-derived growth were constructed using genetic algorithm and artificial neural network (GA-ANN) methods. The statistical parameters of R2 and RMSE are 0.82 and 0.21, respectively which show a considerable improvement compared with multiple linear regression and artificial neural networks (MLR-ANN) method. Ten-fold shuffling cross-validations were carried out to select the most important descriptors. The accuracy of GA-ANN was illustrated by validation techniques of leave-one-out and leave-multiple-out cross-validation and also by Y-randomization. In the second part, Quantitative structure-activity relationship (QSAR) approach was carried out for the prediction of inhibitory activity of some novel quinazolinone derivatives on serotonin (5-HT7) using modified ant colony (ACO) method and adaptive neuro-fuzzy interference system (ANFIS) combined with shuffling cross validation technique. A modified ACO algorithm is utilized to select the most important variables in QSAR modeling and then these variables were used as inputs of ANFIS to predict 5-HT7 receptor binding activities of quinazolinone derivatives. The best descriptors describing the inhibition mechanism are Qmax, Se, Hy, PJI3 and DELS which are electronic, constitutional, geometric and empirical descriptors. The statistical parameters of R2 and root mean square error are 0.775 and 0.360, respectively. In the third part, the inhibitory activity of some pyridine N-oxide derivatives against human severe acute respiratory syndrome (SARS) was predicted in terms of quantitative structure-activity relationship (QSAR) models with the aid of multivariate adaptive regression spline (MARS) and adaptive neuro fuzzy inference system (ANFIS) combined with shuffling cross validation technique. A shuffling MARS algorithm is utilized to select the most important variables in QSAR modeling and then these variables were used as inputs of ANFIS to predict SARS inhibitory activities of pyridine N-oxide derivatives. The statistical parameters of R2 and root mean square error are 0.884 and 0.359, respectively. Comparison of the results of the proposed method with GA-PLS-ANFIS method shows that the shuffling MARS-ANFIS model is superior and can be considered as a tool for predicting the inhibitory behavior of SARS drug-like molecules. In the fourth part, near infrared (NIR) spectroscopy measurements was used on the human plasma of thirty eight patients infected with human immunodeficiency virus type-1 (HIV-1) and twenty uninfected individuals as a control. NIR spectra in the 600-1100 nm region for plasma from HIV-1 infected individuals and healthy donors were subjected for pattern recognition methods. For the classification, genetic algorithm-discriminating partial least squares (GA-DPLS) and counter propagation neural network (CPNN) were employed to visualize the separation between the groups. Raw, first and second derivatives NIR spectra were compared to develop a robust classification. The stability of models was determined by repeatedly splitting the data into training and test sets and computing the percent correctly classified (CC) and predictive ability (PA). Overall, NIR spectroscopy of plasma combined with the pattern recognition was shown to have significant potential as a rapid, accurate and cost-effective tool for diagnosis of HIV-1 infection.
  9. Keywords:
  10. Artificial Neural Network ; Pattern Recognition ; Neuro Fuzzy Network ; Ant Colony Algorithm ; Adaptive Neuro-Fuzzy Inference System (ANFIS) ; Chemometrics Method ; Quantitative Structure-Activity Relationship (QSAR)Model ; Infrared Spectrometry ; Spline Regression ; Drug-Like Molecules

 Digital Object List

  • محتواي پايان نامه
  •   view

 Bookmark

No TOC