Prediction of the Protein O-glycosylation site by Neural Networks and Support Vector Machine

通过机器学习预测哺乳动物蛋白中的o -糖基化位点，以阐明糖基化机制。特别是，将拥挤的糖基化位点和孤立的糖基化位点分开考虑，发现前者受邻近位点氨基酸组成的影响，后者受氨基酸序列的影响，并对其更详细的特征进行了研究。o -糖基化位点经常出现在内在无序的蛋白质中，这导致了对其保护特性、在结构稳定性中的作用等的生物学理解。为了验证预测，还进行了生物实验。

O-Glycosylation sites in the mammalian protein is predicted by the machine learning to elucidate the glycosylation mechanisms. Especially, the crowded and the isolated glycosylation sites are separately considered, and it is found that the former is affected by the amino acid composition of the neighboring sites while the latter is affected by the amino acid sequence, and more detailed characteristics are investigated. O-Glycosylation sites are often found in the intrinsically disordered proteins, which lead to the biological understandings of the conservation property, the role in the structural stability and so on. Biological experiments are also executed for a verification of the prediction.