不同嗜盐机制微生物蛋白质组特性及其识别

选取两种不同嗜盐机制微生物蛋白质组,并将其同非嗜盐微生物蛋白质组进行比较.研究结果发现:积累无机盐的蛋白质氨基酸组成与非嗜盐相差明显,而积累细胞相容性物质的嗜盐蛋白则差异较小,后者分子中His和分子量较小的氨基酸均显著多于非嗜盐蛋白,而Ala则相反;两种类型蛋白质中酸性氨基酸和碱性氨基酸的差值均显著高于非嗜盐蛋白.基于此,使用一种新型Person通用核函数的支持向量机对3种类型蛋白进行识别,其精度可达84.1%,优于其他核函数的支持向量机及其他机器学习算法.

Amino Acid Signatures of Different Hypersanline Adaptation Proteomes and Their Classification

We selected two halophilic proteomes with different halophilic mechanism, and compared with a non-halophilic one. The results showed the difference between the halophilic(salt-in)and the non-halophilic proteome was obvious than that of halophilic(salt-out)and the non-halophilic proteome. In the halophilic(salt-out)proteome, the His and the small residues were significantly higher than those of non-halophilic proteome, while the Ala was significantly lower. However, both halophilic proteomes showed a large excess of acidic over basic amino acids. Based on these results, we introduced a novel Person Universal Kernel Function based support vector machine to classify the three kinds of proteins and the overall prediction accuracy could reach 84.1%. This method outperformed support vector machines based on other usually used kernels and other machine learning algorithms.