賴氨酸甲基化在組蛋白與非組蛋白之辨識與研究

Investigation and identification of lysine methylation on histones and non-histone proteins

鄭字修(Tzu-Hsiu Cheng)

指導教授 ： 李宗夷

元智大學/資訊學院/資訊工程學系/碩士(2012年)

https://doi.org/10.6838/YZU.2012.00130

摘要

蛋白質甲基化是眾多後轉譯修飾中的其中一項，在生物程序中參與著許多重要的過程，其中包括轉錄、訊號傳遞及基因表現調控等。不過不像磷酸化及乙醯化，甲基化直到近幾年才逐漸受到生物學家的重視。目前大部分研究甲基化蛋白質後轉譯修飾都集中在組蛋白上，組蛋白是擁有高度生物保守性的蛋白質，其在各個物種的氨基酸序列上皆有著同源的相似度，但是非組蛋白則與組蛋白相反，完全沒有顯著性可言。然而，先前應用生物資訊方式辨識甲基化位置的研究並沒有將蛋白質區分成組蛋白與非組蛋白來討論。在本篇研究中，我們將訓練資料集區分為三個部份，分別是組蛋白、非組蛋白以及全部(未分類)，利用各種不同的屬性結合支援向量機來辨識賴氨酸甲基化的潛在位置。使用five-fold交叉驗證評估預測的效能，而結果顯示將資料區分為組蛋白與非組蛋白的效能優於全部未分類的資料集。除此之外，我們也使用獨立測試資料來評估訓練資料，其敏感性與特異性的效能在組蛋白為85.62%及80.32%，非組蛋白為69.1%及88.72%。根據以上研究，我們將訓練資料集效能預測最好的屬性，發展成一個有效的系統蛋白質賴氨酸甲基化預測網站(MethK，網站位址:http://140.138.144.141/MethK)。

關鍵字

賴氨酸 ； 甲基化 ； 組蛋白 ； 非組蛋白

並列摘要

Protein methylation is one kind of posttranslational modifications, which was involved in many important biological processes including transcription activity, signal transduction and regulation of gene expression. Biological researchers have not paid as much attention on methylation as on phosphorylation and acetylation until recent years. Post-translational modification of lysine methylation is mostly studied in histones. The histones, which have highly conserved proteins, are homologous in most species; however, non-histone proteins are different from histones that there is no significant motif among the amino acid sequences. Recently, several studies have been proposed that utilizing bioinformatics method to identify protein methylation sites but do not investigate the difference of methylation sites on histones and non-histone proteins. Thus, in this study, we categorize methylation data into three parts: all, histones, and non-hisotne proteins. Then, a support vector machine (SVM) is employed with various types of features for identifying methylated lysines on proteins. Evaluation of predictive performance using five-fold cross-validation indicates that the models trained with histones and non-histone datasets perform better than all dataset. Additionally, the independent testing dataset is used to test effectiveness of the constructed models. For histones methylation, the predictive sensitivity and specificity are 85.62% and 80.32%, respectively. For the methylation of non-histone proteins, the predictive sensitivity and specificity are 69.1% and 88.72%, respectively. Finally, the models yielding best performance are used to implement an effective web server called “MethK” for identifying methylated lysine sites. The MethK is freely available via http://140.138.144.141/MethK.

並列關鍵字

lysine ； methylation ； histones ； non-hitone

參考文獻

1. Paik, W.K. and S. Kim, Enzymatic methylation of protein fractions from calf thymus nuclei. Biochem Biophys Res Commun, 1967. 29(1): p. 14-20.

2. Lee, D.Y., et al., Role of protein methylation in regulation of transcription. Endocr Rev, 2005. 26(2): p. 147-70.

3. Springer, M.S., M.F. Goy, and J. Adler, Protein methylation in behavioural control mechanisms and in signal transduction. Nature, 1979. 280(5720): p. 279-84.

4. Stallcup, M.R., Role of protein methylation in chromatin remodeling and transcriptional regulation. Oncogene, 2001. 20(24): p. 3014-20.

5. Clarke, S., Protein methylation. Curr Opin Cell Biol, 1993. 5(6): p. 977-83.