使用類神經網路預測蛋白質結構; Application of Neural Networks in Predicting Protein Structures

NCUIR > College of Electrical Engineering & Computer Science > Executive Master of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/8680

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/8680

Title:	使用類神經網路預測蛋白質結構;Application of Neural Networks in Predicting Protein Structures
Authors:	謝博文;Po-Wen Hsieh
Contributors:	資訊工程學系碩士在職專班
Keywords:	蛋白質結構預測;倒傳遞;Prediction of Protein Structure;Back Propagation
Date:	2005-07-04
Issue Date:	2009-09-22 11:32:52 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	蛋白質三級結構的定序方式，目前約分兩種─核磁共振(NMR)與X射線(X-ray)。但由於此兩種定序方法成本高且耗時，因此發展出同源模擬法(Homology modeling)、摺疊辨識法(Fold recognition)、重頭起算法(ab initio)來預測蛋白質分子結構。此類研究的技術可以預測出蛋白質結構的模板(template)，預測結果可以應用在蛋白質突變(protein mutation)研究，活性位置研究(active site)，藥物設計等，並能減少在實驗上、製藥上所須的時間。本文將對蛋白質序列與蛋白質結構，以類神經網路(Neural Networks)之倒傳遞演算法(Back Propagation)，來訓練有可靠的NMR及X-ray鑑定的蛋白質，一級序列與三級結構的關係，其結果可預測未知的一級蛋白質序列的三級結構。 Presently, there are two methods for sequence of three-dimensional protein structure, they are NMR (Nuclear Magnetic Resonance) and X-ray. Since these two methods cost high charge and time, the Homology modeling, Fold recognition and ab inition methods were developed to predict the structure of protein molecule. These techniques can predict the template of protein structure and their prediction results can be applied in the research of protein mutation, active site, medicine design, etc. Also, these techniques can reduce the development time for experiment and pharmacy. Therefore, the reliable data of protein 1D sequence obtained by NMR and X-ray are as the training patterns in our prediction model. For our study, we use back propagation algorithm of neural network to do simulation to project the 3D protein template through its 1D sequence.
Appears in Collections:	[Executive Master of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...