论文部分内容阅读
对用于折叠模式识别的蛋白质结构数据库进行结构分类,构建了四个分类库:Al-α库,Al-β库,α/β库,α+β库和一个总库,然后分别统计出不同灵敏度的匹配评估函数(平均势)。对不同的平均势,不同结构类型的蛋白进行的检验发现:来源于α/β库的平均势预测能力最强,来源于Al-α库的平均势预测能力最弱;对α/β蛋白的预测成功率最高,对Al-α蛋白的预测成功率最低。这与α/β蛋白结构最规则,Al-α蛋白未加入辅基不能反映出结构的全部特征是相一致的
The structural classification of protein structure database used for folding pattern recognition was classified, and four classification libraries were constructed: Al-α, Al-β, α / β, α + β and a total database. Matching evaluation function (average potential). The results of different mean potential and different structural types of proteins showed that the average potential potential derived from the α / β bank was the strongest, the average potential potential derived from the Al-α bank was the weakest, and the α / β protein The highest prediction success rate is the lowest prediction success rate for Al-α protein. This is consistent with the fact that the most regular structure of the α / β protein, the absence of the Al-α protein in the prosthetic group, does not reflect all of the structural features