基于序列特征预测先天性糖基化紊乱疾病基因

Predicting disease genes inducing congenital disorders of glycosylation using sequence features

选取mRNA和protein序列的330个物化特征进行研究,发现某些特征在已知的先天性糖基化紊乱(CDG)疾病基因中具有共同显著性差异.基于这个特点设计了一种新的CDG疾病基因预测方法,留一法交叉验证结果表明该方法能够有效地从众多候选基因中预测出真实的CDG基因,效果略优于其他预测方法.本方法只需要知道mRNA和protein的序列数据,弥补了目前蛋白质功能、相互作用等数据不足或不可靠的缺陷.进一步对其他疾病基因的预测结果表明本方法具有一定的推广性.

Chosed 330 physical and chemic features of mRNA and protein, and found certain of them show similar significant differences on known disease genes of congenital disorders of glycosylation （CDG）. This property was used to develop a novel approach to predicting CDG disease genes. The results of leave-one-out cross-validation demonstrated that this approach could slightly detect actual CDG disease;genes from numbers of candidate genes, and its effect exceeds other approaches＇. This approach is merely based on mRNA and protein sequences, remedied the present existing defect caused by incomplete or unreliable data of protein functions and interactions and so on. This approach is used to predict other diseases＇ genes and the results indicate that it has certain applicability