磷酸化是蛋白質中最重要的後轉譯修飾之一,它參與各種不同的生物訊號傳遞的路徑。在傳遞路徑中,蛋白質的磷酸化與去磷酸化反應為真核細胞提供了調控機制,控制了許多酵素和受體的活化與抑制。因此,若能預測磷酸化作用的位置,將對相關的研究有很大的助益。 本論文提出了一個新的序列保留性計算法則,建立起一個模型,它是利用已知磷酸化的片段當作預測的模板,並利用其與非磷酸化的片段比對,求出最大相似度,並依此建立預測磷酸化的門檻值,最後將未知待預測序列的片段送入本預測模型。實驗結果並與Netphos、Disphos和GPS三個網站比較,針對五個性質來做探討:靈敏度(Sn)、特異性(Sp)、準確度(Acc)、精準度(Precision)和馬修斯相關係數(MCC)。實驗預測的結果顯示,本方法在馬修斯相關係數及預測準確度都有相當不錯的表現。
Phosphorylation, one of the most important post-translational modifications in proteins, is involved in biological signal transduction in various processes. During signal transduction, proteins’ phosphorylation and dephosphorylation provides a control mechanism for eukaryotes to activate or inhibit a number of enzymes and receptors. Therefore, if we can predict where phosphorylation is taking place, it will benefit relevant research significantly. In this study, a new sequence conservation method is proposed to create a model. Using known phosphorylated segments as a template for prediction and matching them with non-phosphorylated segments, we attempted to find their maximum similarity. Based on this, a threshold value for prediction was then for each template. Finally, unknown sequence segments to be predicted were substituted into the prediction model. The results were then compared with the 3 websites Netphos, Disphos and GPS, and a discussion on the following 5 kinds of estimation was conducted accordingly: Sensitivity (Sn), Specificity (Sp), Accuracy (Acc), Precision, and Mattews correlation coefficient (MCC). The results show that this method had relatively good performance in MCC and Acc prediction.