Zhou X-B, Chen C, Li Z-C, Zou X-Y
School of Chemistry and Chemical Engineering, Sun Yat-Sen University, Guangzhou, China.
Amino Acids. 2008 Aug;35(2):383-8. doi: 10.1007/s00726-007-0608-y. Epub 2007 Dec 21.
Apoptosis proteins play an important role in the development and homeostasis of an organism. The accurate prediction of subcellular location for apoptosis proteins is very helpful for understanding the mechanism of apoptosis and their biological functions. However, most of the existing predictive methods are designed by utilizing a single classifier, which would limit the further improvement of their performances. In this paper, a novel predictive method, which is essentially a multi-classifier system, has been proposed by combing a dual-layer support vector machine (SVM) with multiple compositions including amino acid composition (AAC), dipeptide composition (DPC) and amphiphilic pseudo amino acid composition (Am-Pse-AAC). As a demonstration, the predictive performance of our method was evaluated on two datasets of apoptosis proteins, involving the standard dataset ZD98 generated by Zhou and Doctor, and a larger dataset ZW225 generated by Zhang et al. With the jackknife test, the overall accuracies of our method on the two datasets reach 94.90% and 88.44%, respectively. The promising results indicate that our method can be a complementary tool for the prediction of subcellular location.
凋亡蛋白在生物体的发育和体内平衡中起着重要作用。准确预测凋亡蛋白的亚细胞定位对于理解凋亡机制及其生物学功能非常有帮助。然而，现有的大多数预测方法都是利用单一分类器设计的，这将限制其性能的进一步提高。本文提出了一种新颖的预测方法，该方法本质上是一个多分类器系统，它将双层支持向量机（SVM）与包括氨基酸组成（AAC）、二肽组成（DPC）和两亲性伪氨基酸组成（Am-Pse-AAC）在内的多种组成相结合。作为示例，我们在两个凋亡蛋白数据集上评估了该方法的预测性能，这两个数据集分别是由周和多克托生成的标准数据集ZD98，以及由张等人生成的更大的数据集ZW225。通过留一法检验，我们的方法在这两个数据集上的总体准确率分别达到了94.90%和88.44%。这些令人鼓舞的结果表明，我们的方法可以作为预测亚细胞定位的补充工具。
