利用支持向量機與機率矩陣以序列預測蛋白質與蛋白質交互作用之接合點 ; Sequence-based prediction of protein-protein interaction sites using probability matrix and support vector machine

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/44717

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/44717

Title:	利用支持向量機與機率矩陣以序列預測蛋白質與蛋白質交互作用之接合點;Sequence-based prediction of protein-protein interaction sites using probability matrix and support vector machine
Authors:	朱致遠;Chih-Yuan Chu
Contributors:	資訊工程研究所
Keywords:	蛋白質與蛋白質交互作用接合點;protein-protein interaction sites
Date:	2010-07-27
Issue Date:	2010-12-09 13:53:34 (UTC+8)
Publisher:	國立中央大學
Abstract:	蛋白質與蛋白質交互作用在生物功能上扮演了很重要的角色，而分析蛋白質與蛋白質交互作用時的接合點是一種研究蛋白質與蛋白質交互作用的普遍方法。此外對蛋白質的接合研究以及藥物設計等等都有很大的幫助。胺基酸殘基的組成對一個殘基是否成為一個接很點有很大的影響力，我們嘗試蒐集了大量的胺基酸組合資料以作為預測之用。我們提出這個方法的基本概念就是當殘基在特定組合中過去偏向於有接合點發生時，在未來於相同組合中應也偏向於發生接合，反之亦然。我們和其他三篇研究作出比較，並且都顯示出我們的方法有較高的準確率。而平均靈敏度以及特異度都接近九成。雖然我們的結果較好，但我們的方法可以說是暴力解，在利用支持向量機前，我們先預先收集並統計了萬筆的蛋白質序列資料。若忽視我們的方法可能需要大量的電腦記憶體，我們提供了一個高準確率的蛋白質與蛋白質交互作用接合點的預測方法。Protein-protein interaction plays a key role in many biological functions. Identifying protein-protein interaction site is a common way to understand protein-protein interaction and is helpful in docking studies, drug design and so on. Since amino acid residue combination has a great related to interaction site forming, we tried to collect and calculate large amount of combination for predicting usage. The basic idea in our method is residues in certain combinations will prefer interaction if it acted as interaction sites more often before in same combinations, vice versa. We compared our method with other three studies and performed better in all cases. The average of sensitivity and specificity were both nearly 90%. Although our performance was better than others, our method was nearly a brute force since we collect and calculate nearly ten thousand protein sequences before. If ignore the fact that our method may needed large computer memory, we provide a highly accuracy method for protein-protein interaction site prediction.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	547	View/Open

社群 sharing

Loading...