蛋白質結合殘基預測之研究

A Study of Prediction of Protein Binding Residues

黃鈺峰(Yu-Feng Huang)

指導教授 ： 歐陽彥正

共同指導教授 ： 黃乾綱

國立臺灣大學/電機資訊學院/資訊工程學研究所/博士(2010年)

https://doi.org/10.6342/NTU.2010.10695

摘要

蛋白質功能性區塊的探索一直以來都是生物學家所追求的。蛋白質透過自身的殘基與相互作用的對象進行作用時，作用區內的殘基扮演著不同的角色，有些是負責進行核心作用的辨識機制，而有些則是負責穩定蛋白質與互動對象間的結合環境。因此，預測蛋白質結合殘基座落的位置仍然是生物學家所關心的重大議題，不論是透過序列或結構資訊的深入分析都是重要且必須的處理方式。
對於蛋白質結合殘基預測的方法主要可以分成兩大類：序列為主與結構為主。本論文分別提出序列與結構兩個面向的架構來解決，並分別應用在轉錄因子以及酵素上。透過蛋白質序列來預測轉錄因子上專一性殘基與非專一性參級的座落位置，並且經由結合殘基預測的資料更進一步預測蛋白質與去氧核醣核酸間的結合模式。透過結合模式的預測，我們可以有效地增加預測的效能。而在酵素這類型的蛋白質，雖然這類型的蛋白質是屬於相對穩定的蛋白質，但由於配體的形變因而也造成結合區塊隨之改變。因此我們透過蛋白質穩定區塊所探勘而成的結構模板藉由預測酵素家族來檢定所探勘之結構模板的品質。
最後，對於一個未知的蛋白質，藉由預測的方式取得可能的結合殘基位置可以協助生物學家進行較為精準的生化實驗。而實驗結果也呈現出蛋白質與不同相互作用的對象具有不同的功能，結合環境和特異性。因此藉由更細緻的分群方式使可以有效提升預測的準確性。

關鍵字

機器學習 ； 資料探勘 ； 蛋白質作用區(結合殘基)預測 ； 轉錄因子 ； 與DNA作用之蛋白質 ； 酵素

並列摘要

Protein carries out its function by interaction between binding sites/residues and its interacting partners. Functional important residues recognize and bind with protein’s interacting partners; residues close to functional important residues play roles to support functional site’s activity; remaining residues play roles to construct global protein structure. Therefore, prediction of protein binding residues from sequence and/or structure is investigated deeply and still a great challenge for many years.
Two directions to predict protein binding sites and then infer protein functions are methods based on evolutionary information and methods based on structural template library. From protein sequence information, predicting DNA-binding residues in transcription factors based on machine learning approach with evolutionary information can help biologists for further experiments because transcription factors are mostly disorder proteins. From protein structure information, structural templates are extracted by stable region identification for identifying enzyme family without the help of protein-ligand complexes.
In conclusion, predicting protein binding residues and protein functions is the first step to help biologists to have rudimentary view of proteins with annotations. Experimental results reveal that proteins binding with different interacting targets have different binding environment and specificity. Grouping proteins by protein mechanism, binding target, or structural conformation for predicting binding residues is suggested as well.

並列關鍵字

machine learning ； data mining ； protein binding sites/residues prediction ； transcription factor ； DNA-binding protein ； enzyme.

參考文獻

1. Crick, F., Central dogma of molecular biology. Nature, 1970. 227(5258): p. 561-3.

4. Jones, S., et al., Protein-RNA interactions: a structural analysis. Nucleic Acids Res, 2001. 29(4): p. 943-54.

5. Jones, S., et al., Protein-DNA interactions: A structural analysis. J Mol Biol, 1999. 287(5): p. 877-96.

6. Mirny, L.A. and M.S. Gelfand, Structural analysis of conserved base pairs in protein-DNA complexes. Nucleic Acids Res, 2002. 30(7): p. 1704-11.

7. Bartlett, G.J., et al., Analysis of catalytic residues in enzyme active sites. J Mol Biol, 2002. 324(1): p. 105-21.