VARIABLE SELECTION METHOD IMPROVES THE PREDICTION OF PROTEIN SECONDARY STRUCTURE FROM CIRCULAR-DICHROISM SPECTRA
VARIABLE SELECTION METHOD IMPROVES THE PREDICTION OF PROTEIN SECONDARY STRUCTURE FROM CIRCULAR-DICHROISM SPECTRA
复制标题
DOI：
10.1016/0003-2697(87)90135-7
复制
发表时间：
1987-11-15
影响因子：
2.9
通讯作者：
JOHNSON, WC
中科院分区：
文献类型：
--
作者：
MANAVALAN, P;JOHNSON, WC
本文提出了一种基于“变量选择”统计方法的蛋白质圆二色谱二级结构预测方法。变量选择增加了Provencher和Glockner方法中的灵活性（S。W. Provencher和J. Glockner，1981，Biochemistry 20，33-37）至Hennessey和约翰逊的方法（J. P. Hennessey和W. C.约翰逊，1981，生物化学20，1085-1094）。提出了两种分析方法，用于从由Provencher和Glockner方法产生的系列中选择解决方案，这改进了该技术。结果表明，变量选择法和改进的Provencher和Glockner法的可靠性均上级原Hennessey和约翰逊法。对于新的变量选择方法，对于测量到178 nm的数据，在X射线结构和预测的二级结构之间计算的相关系数为：螺旋，0.75对于β-薄片，β-0.50转弯，其他结构为0.89。虽然变量选择方法改善了在190 nm处截断的圆二色性数据的分析，但测量到178 nm的数据给出了上级结果。它表明，提高拟合到测量CD超出数据的准确性，可能会导致较差的分析。
A new procedure based on the statistical method of "variable selection" is used to predict the secondary structure of proteins from circular dichroism spectra. Variable selection adds the flexibility found in the Provencher and Glockner method (S. W. Provencher and J. Glockner, 1981, Biochemistry 20, 33-37) to the method of Hennessey and Johnson (J. P. Hennessey and W. C. Johnson, 1981, Biochemistry 20, 1085-1094). Two analytical methods are presented for choosing a solution from the series generated by the Provencher and Glockner method, and this improves the technique. All three methods are compared and it is shown that both the variable selection method and the improved Provencher and Glockner methods have equivalent reliability superior to the original Hennessey and Johnson method. For the new variable selection method, correlation coefficients calculated between X-ray structure and predicted secondary structures for data measured to 178 nm are: 0.97 for .alpha.-helix, 0.75 for .beta.-sheet, 0.50 for .beta.-turn, and 0.89 for other structures. Although the variable selection method improves the analysis of circular dichroism data truncated at 190 nm, data measured to 178 nm gives superior results. It is shown that improving the fit to the measured CD beyond the accuracy of the data can result in poorer analyses.