蛋白质折叠规律研究是生命科学重大前沿课题,折叠类型分类是蛋白质折叠研究的基础。构建BRD—like折叠类型模板数据库,建立了基于多模板的综合分类方法,并用于该折叠类型的分类。对实验集的12117个样本进行检验,结果的敏感性、特异性分别为0.923和0.997,MCC值为0.72;对独立检验集2260个样本的检验,结果发现:敏感性、特异性分别为0.941和0.998,MCC值为0.86.结果表明:基于多模板的综合分类方法可用于蛋白质折叠类型分类。
The study on principle of protein folding is a cutting-edge topic in life science, and folding type classification is the basis of protein folding research. In this paper, we constructed a template database of BRD-like folding type, and established a comprehensive classification method based on multiple templates. Our method is used for the classification of BRD-like folding. We tested the training set of 12 117 samples,and found that the sensitivity,specificity and MCC were 0.923,0.997 and 0.72 respectively.Then we tested the 2 260 samples of the independent test, and found that the sensitivity, specificity and MCC were 0.941,0.998 and 0.86 respectively. These results indicated that the comprehensive classification method based on multiple templates could be used for the classification of protein folding.