Prediction of late embryogenesis abundant proteins with chaos games representations

摘要

Late embryogenesis abundant proteins (LEA) play vital role in rice breeding. According to the sequence statistical nature of this protein family, sequence-based approach could be used in LEA protein classification and prediction. By using a novel chaos games representation method, efficient feature extraction algorithm is obtained to reflect the intrinsic quality of given proteins. With training and testing through support vector machine, the best classifier achieves a high accuracy of 93.37%, and Mathews correlation coefficient reaches 0.8690.