基于多类特征融合方法预测细胞凋亡蛋白质的亚细胞位置

Predicting the Subcellular Location of Apoptosis Proteins Based on Multi-features Fusion

细胞凋亡蛋白质在调控细胞死亡与增殖问的平衡方面起着至关重要的作用，而其生物学功能与亚细胞位置有着紧密的联系，因此对凋亡蛋白质亚细胞定位的研究有助于对其功能做进一步的了解。文中利用了蛋白质骨架信息、平均化学位移信息、氨基酸，2肽组分和序列亲疏水分布等信息，并基于多类特征融合的方法，采用支持向量机（SVM）算法对我们新构建的细胞凋亡蛋白质亚细胞定位数据集进行了分类预测，在Jackknife检验下，总的预测成功率达到了80．2％，均高于单个特征信息得到的总体预测成功率，这一结论说明特征融合的方法可以有效地应用到细胞凋亡蛋白质亚细胞位置预测的研究中。

Apoptosis proteins are crucial for regulating the balance between cell death and renewal. The biological functions of an apoptosis protein are closely related to its subcellular location in a cell. So, predicting the subcellular location of apoptosis proteins will help us understand the biological functions of the apoptosis proteins better. Several biological features, protein blocks composition, average chemical shifts composition, amino acid n-peptide composition information and the hydropathy distribution along protein sequence, were effectively applied to predict the subcellular location of apoptosis protein by using support vector machine （SVM） algorithm. The overall prediction accuracies of the jack knife tests based on the fused feature information is 80.2%, which is higher than another feature. The results show that the approach by multi-features fusion is pretty useful for predicting apoptosis protein＇s subcellular location.