經由辦識作用樣式預測蛋白質交互作用

Protein-Protein Interaction Prediction with Identification of Putative Interaction Patterns

蔡孟家(Meng-Jia Tsai)

指導教授 ： 高成炎

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2007年)

https://doi.org/10.6342/NTU.2007.01588

摘要

蛋白質交互用作與其作用位置是了解生物體內運作的重要資訊，現在雖然已發展了許多實驗的方法來加以探測，但能得到的結果仍只是自然界的極小部分，為了增加探索速度，不少研究人員開發計算性的方法來預測，交互作用預測的準確度大致上都可以接受，但不是每個方法都能辨識出交互作用發生的位置，而一般用來預測作用介面的方法大都需要有三級結構的資訊來輔助，可是相較於已經解出的基因體序列，已知的結構資訊實在是太少了，所以這類的方法並不見得適用。

我們提出一個資料探勘演算法，其修改自挖掘關聯規則的Apriori演算法，可從已知的一小群蛋白質交互作用，再配合對應的一級結構序列，辨認出潛在的交互作用片段樣式，利用這些結果可用來預測未知的交互作用，以及分析可能的交互作用位置。利用三級結構來驗證，有些樣式確實能看出其關聯性。另外，用來做大規模預測的準確度雖只有~65%，不過在高可信度的預測卻可達~85%，所以在實際的應用上，應有相當的價值。

關鍵字

預測蛋白質交互作用 ； 預測蛋白質交互作用位置 ； Apriori演算法 ； 序列基礎的預測 ； 高可信度的預測

並列摘要

Protein-protein interaction and interaction site information are key points to realize biological processes. Experimental approaches for detecting protein-protein interactions are far behind the tremendous number of possible interactions. Computational approaches for predicting protein-protein interactions are able to have reasonable accuracies but are usually not able to identify interaction sites. Currently, many methods based on three-dimensional (3D) structures are capable of elucidating interaction interfaces. However, the 3D structure-based methods are computationally intensive and rely on the structures of proteins which may not be readily available. Features and patterns from primary structure can be applied to predict interactions, even potential interaction sites. Therefore, we propose an Apriori-like algorithm to identify possibly interaction-related pairs of short peptides from protein sequence and known interactions. Subsequently, the discovered pairs are able to predict interactions and to infer interaction sites. We concentrate on tri-gram interaction- related and non-interaction-related peptide pairs in the different species, and obtain acceptable performance for high-throughput and high-confidence predictions, with accuracies of ~65% and ~85%, respectively. Moreover, in the 207 hetero-sequenced pairs of chains with 3D structures, some discovered pairs are highlighted in their respective 3D structures. The results presented suggest that these pairs are correlated to binding sites and reasonable for interaction prediction.

並列關鍵字

Protein-protein interaction prediction ； Protein-protein interaction site prediction ； Apriori-like algorithm ； Sequence-based prediction method ； High-confidence prediction system

參考文獻

Appling,D.R. (1999) Genetic approaches to the study of protein-protein interaction. METHODS, 19, 338–349.

Bader,G.D., Donaldson,I., Wolting,C., Ouellette,B.F., Pawson,T., and Hogue,C.W. (2001) BIND—the biomolecular interaction network database. Nucleic Acids Res., 29, 242–245.

Baldi,P., Brunak,S., Chauvin,Y., Andersen,C.A.F., and Nielsen,H. (2000) Assessing the accuracy of prediction algorithms for classification: an overview. Bioinformatics, 16, 412–424.

Ben-Hur,A. and Noble,W.S. (2005) Kernel methods for predicting protein-protein interactions. Bioinformatics, 21, i38–i46.

Bender,A. and Pringle,J.R. (1991) Use of a screen for synthetic lethal and multicopy suppressee mutants to identify two new genes involved in morphogenesis in Saccharomyces cerevisiae. Mol Cell Biol., 11, 1295–1305.