A Fast and Effective Approach of Fold Recognition Based on Image Feature
使用图像特征构建快速有效的蛋白质折叠识别方法

One of the most important research aims is to understand the relationship between structure and function of protein. Inspired by this aim, automatic classification of protein structure becomes one of major research approaches. However, how to extract compact and effective feature to characterize protein structure is still a challenge to it. In this paper, 3-D tertiary structure of protein fold was mapped into 2-D distance matrix which can be further regarded as gray level image. Next, based on histogram and gray level co-occurrence matrix (CoM), a feature extraction of fold structure with low-cost computation was presented and feature vector with low dimension and definite structural properties was obtained. Furthermore, the nature of histogram peak at gray level 0 was depicted, and the structural meanings of gray distribution, various angles and pixels distance of CoM were analyzed in detail respectively. Finally, the presented feature extraction was validated by classification of 27 types of folds, and compared with several feature methods based on sequence or structure. The presented method achieved the accuracy 71.95% in independent test by using 5-CV (cross validation) to select the parameters of support vector machines (SVM), and 78.94% with 10-CV test on the whole combined data of training and testing sets. The results show that the presented method can perform effectively and efficiently automatic classification of multiple types of folds with the benefit of low dimension and compact feature, but also no need of complicated classifier system.