论文部分内容阅读
本研究采用偏最小二乘法(PLS)建立了抗类风湿性免疫活性肽的QSAR模型。首先从常见氨基酸各原子间静电、立体和疏水等与生物活性直接相关的非键作用方式,筛选出适合于免疫活性肽定量-构效建模的氨基酸描述符,再采用此描述符对47个肽数据集进行定量构效关系建模。结果表明,Z-scales氨基酸描述符实现了对氨基酸残基的多位置、多变量的定量结构描述,最适合用于描述免疫活性肽的物化性质。根据Z-scales描述符,经主成分分析和偏最小二乘法建立的模型有良好的可靠性和预测能力,模型的复相关系数R2=0.986,均方根误差RMSE=0.253,留一交叉验证相关系数Q2LOO=0.893,外部验证相关系数Q2ext=0.971。通过对免疫活性肽构效关系的研究为类风湿性关节炎新耐受原的筛选、设计开辟了广阔空间,同时也为功能食品的开发和创新提供了新手段。
In this study, QSAR model of anti-rheumatic immunologically active peptide was established by partial least squares (PLS). Firstly, the amino acid descriptors suitable for the quantitative-structure-property modeling of immunologically active peptides were screened out from the non-bonding mode which is directly related to the biological activity of common amino acids such as electrostatic, stereo and hydrophobic, and then 47 Peptide data sets for QSAR modeling. The results show that the Z-scales amino acid descriptors enable multi-position and multivariate quantitative structure elucidation of amino acid residues and are most suitable for describing the physicochemical properties of immunologically active peptides. According to the Z-scales descriptor, the model established by principal component analysis and partial least squares has good reliability and predictability. The complex correlation coefficient R2 = 0.986 and root mean square error RMSE = 0.253, with a cross-validation correlation Coefficient Q2LOO = 0.893, external verify correlation coefficient Q2ext = 0.971. The study of the structure-activity relationship of immunologically active peptides has opened up a broad space for the screening and design of new tolerogenic agents for rheumatoid arthritis, and also provided a new method for the development and innovation of functional foods.