以蛋白質序列物化特性為特徵的蛋白質激Kinase-Specific磷酸化位置預測方法與分析

Abstract

本計畫「以蛋白質序列物化特性為特徵的蛋白質激酶磷酸化位置預測方法與分析」旨在提 出一套高效能的智慧型基因演算法來擷取蛋白質序列物化特性做為特徵，例用支持向量迴歸方 法來預測蛋白質相對溶劑可接觸性(RSA,代表蛋白質上某一氨基酸和溶劑接觸程度)，並例用支 持向量機與物化特性及RSA 預測值來預測蛋白質激酶磷酸化位置。一般結合處是在蛋白質的 表面，因此氨基酸位於表面程度與氨基酸的物化特性有相關，也是預測蛋白質殘基溶劑可接觸 性的重要特性。本計畫提出一個二階段探勘與分析序列物化特性來設計利用溶劑可接觸性面積 輔助預測磷酸化發生位置的方法。由於蛋白質序列物化特性的可解讀性高，故可分析RSA 與 磷酸化的物化特性差異性。本計劃將特別以實驗室驗證磷酸化發生的位置的蛋白質序列-瞬時 感受電位第四號TRPV4 做為分析標的物，作一預測系統以提供驗證。

The one-year project, prediction and analysis of identifying protein kinase-specific phosphorylation sites based on the features of physicochemical properties of sequences, aims to develop a set of high-performance intelligent genetic algorithms for mining physicochemical properties of sequences as features to design two prediction methods. The first method uses support vector regression with the physicochemical features to predict relative surface area (RSA) of solvent accessibility. The second method uses support vector machine with the physicochemical features to predict protein kinase-specific phosphorylation sites. The RSA value plays an important role in developing explicit models for aiding prediction of the phosphorylation sites. Therefore, the project proposes a two-step prediction method using the predicted RSA value of step 1 as an additional important feature to predict phosphorylation sites. Due to the high interpretability of physicochemical properties as features, the study will analyze the similarity and difference of physicochemical properties to understand the mechanism of phosphorylation. Furthermore, the project would develop a prediction platform to verify known proteins with phosphorylation sites, such as transient receptor potential vanilloid subfamily member 4 (TRPV4).