论文部分内容阅读
从分子的三维空间结构出发,按照不同类型原子之间静电和立体作用得到一种新的分子结构表达方法——三维原子场相互作用矢量(3D-VAIF).利用该矢量对20种天然氨基酸空间结构性质进行计算,通过主成分分析获得单个氨基酸的特征描述子——氨基酸结构信息得分(SSIA).分别使用58个血管紧张素转化酶抑制剂、48个苦味二肽以及31个缓激肽对SSIA的性能进行了测试,所得模型的复相关系数R2cum和交互检验复相关系数Q2LOO分别为0.789,0.773;0.856,0.837和0.838,0.815.结果表明将SSIA用于肽类似物定量序列活性建模(定量序效建模,QSAM)效果优于传统氨基酸描述子.
Starting from the three-dimensional structure of molecules, a new molecular structure expression method, 3D-VAIF, is derived according to the electrostatic and stereoscopic interactions between different types of atoms. Using the vector, 20 natural amino acid spaces Structural properties were calculated, and a single amino acid characteristic descriptor - amino acid structure information score (SSIA) was obtained by principal component analysis. Fifty-eight angiotensin converting enzyme inhibitors, 48 bitter dipeptides and 31 bradykinin pairs SSIA was tested and the correlation coefficients R2cum and Q2LOO of the model were 0.789,0.773, 0.856,0.837 and 0.838,0.815 respectively.The results showed that SSIA could be used to model the activity of peptide analogs QSAM) is superior to traditional amino acid descriptors.