Prediction of Peptide Maps in CZE and MEKC Systems

No Thumbnail Available

Date

2005-04-24

Journal Title

Series/Report No.

Journal ISSN

Volume Title

Publisher

Abstract

A new Quantitative Structure-Migration Relationships(QSMR) model was developed to predict the electrophoretic mobilities of peptides in capillary zone electrophoresis(CZE). A three-step strategy was used: first, select the best charge-size term from the existing models; second, develop a muilti-linear regression(MLR) model to study the linear characteristics of peptide mobility using the best charge-size term and other descriptors; third, generate an artificial neural network(ANN) to investigate the nonlinear behavior of peptide mobility and use this ANN model to predict peptide migration behavior in CZE. To test the robustness of the QSMR model, it was applied to the data published by another research group. Very accurate predictions were achieved. To study the influence of peptide sequence on the migration of a peptide in CZE, a series of 'sequence-related' descriptors were developed. These descriptors were used to develop MLR models for peptide mobility prediction. With the 'sequence-related' descriptor, more accurate mobility could be predicted for peptides with same amino acid composition but different sequences. Group contribution approach(GCA) was used to determine the individual contribution of each amino acid residue and both N-, C- terminal to the peptide mobility in Tween20 system. Data of a relatively small number of peptides were used for this purpose. The sum of individual contributions was calculated for each peptide and used as a new descriptor in developing MLR models for the prediction of peptide mobilities in Tween20 system. Good preliminary results were achieved.

Description

Keywords

peptide mapping separation prediction

Citation

Degree

MS

Discipline

Chemistry

Collections