結構特徵索引設計和索引式蛋白質結構比對演算法的研究

Study on structural signature design and algorithms for index-based protein structure comparison

楊逸偉(Yi-Wei Yang)

指導教授 ： 黃乾綱

國立臺灣大學/工學院/工程科學及海洋工程學研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.02711

摘要

隨著蛋白質資料量的快速增加，如何有效地分析蛋白質之間的關係成為現今非常重要的議題，因此，生物學家可以透過蛋白質結構的分析快速地了解蛋白質之間的反應關係與演化上的關聯性。然而，隨著蛋白質結構資料不斷地以指數般的成長，單純只使用傳統的蛋白質結構比對演算法，來進行蛋白質結構上的資料探勘，如structure motif detection 或是 structure conservation discovery時，將因大量的次結構比對造成長時間耗時的計算而無法應付。因此為加速蛋白質結構比對的速度並運用於蛋白質結構資料探勘上，本研究透過採用索引的技巧，進行蛋白質的局部環境結構編碼，轉換成具有局部結構資訊的一維結構特徵 (structural signature)。並透過索引技巧避免傳統蛋白質結構比對 (protein structure comparison, PSC)所遭遇到的問題，提供同時比對大量蛋白質結構的功效。
本論文透過索引值的設計，將原本複雜三維空間結構，在可容忍的資料流失下轉換成為一維的結構特徵，所設計出來的結構特徵仍保有空間中的資訊。並利用索引及近似比對技巧，在尚未進入蛋白質比對階段前，先快速搜尋相似程度較高的部份結構，有了這個先決條件，可以有效的減少所需比對時間，最後並嘗試帶入原有結構特徵值資訊，找出對應蛋白質之間的共同座標系轉換資訊，進而達到蛋白質結構搜尋比對效用。在實驗部分，本論文使用了蛋白質結構資料庫 (Protein Data Bank)中的資訊來進行驗證，並將比對出的結果參考Enzyme Classification中的資料，並比較VAST和SSM這兩個演算法的效果，驗證所找尋到的蛋白質具有相同功能。就目前所觀察到的實驗數據證明這樣的架構是可行的。藉由本研究所提出的蛋白質結構比對方法，可以讓生物學家在蛋白質結構分析上有更好的工具。

關鍵字

蛋白質結構搜尋 ； 蛋白質結構比對 ； 索引值設計

並列摘要

With the fast growth of determined protein structures in Protein Data Bank, to effectively analyze the relationship between protein structures has become a critical issue. Currently, protein structure comparison is the major approach to analyze protein structures and discover local functional regions. By evaluating similarity among protein structures, we can identify common substructures shared by the group of proteins. However, if we would like to mine local conserved regions among the group of protein structures, pair-wise protein structure comparison could be applied but it is still a time-consumed computational task because of the exponential growth of protein structures.
In this paper, we use sphere-based representation to describe protein local structure and apply indexing technique on protein structure to approach database search on Protein Data Bank for the purpose of protein structure mining. Our approach tries to filter out lots of non-similar protein structures to avoid unnecessary comparisons, and then compares multiple protein structures at the same time. While comparing with VAST and SSM in the capability of find protein structure under hierarchical functional classification of Enzyme Classification, our proposed approach has better performance in sensitivity and precision than VAST and SSM. Our experiment results reveal that our method could be a good approach to find protein structure with similar function. We provide the alternative for molecular biologists to search whole Protein Data Bank from the view of protein local structure.

並列關鍵字

NRS ； indexing ； Protein structure

參考文獻

1. Berman, H.M., et al., The Protein Data Bank. Nucleic Acids Res, 2000. 28(1): p. 235-42.

2. Brown, N.P., C.A. Orengo, and W.R. Taylor, A protein structure comparison methodology. Computers & Chemistry, 1996. 20(3): p. 359-380.

3. Gibrat, J.F., T. Madej, and S.H. Bryant, Surprising similarities in structure comparison. Curr Opin Struct Biol, 1996. 6(3): p. 377-85.

4. Kohel, K.A., et al., Seasonality in plasma thyroxine in the desert tortoise, Gopherus agassizii. Gen Comp Endocrinol, 2001. 121(2): p. 214-22.

5. Holm, L. and C. Sander, Protein structure comparison by alignment of distance matrices. J Mol Biol, 1993. 233(1): p. 123-38.