Efficient Protein Classification by using 3D Structure Content Representation

K. Trivodaliev, S. Kalajdziski, A. Kulakov, D. Davcev, and G. Mirceva (Macedonia)


Protein classification, neural networks, decision trees


In this paper, a 3D structure-based approach is presented for the efficient classification of protein molecules. The method relies on the geometric 3D structure of the proteins. After proper positioning of the 3D structures, the Spherical Trace Transform is applied to them to produce geometry - based descriptors, which are completely rotation invariant. Additionally, some biologi cal properties of the protein are taken, and added to the geometry-based descriptor, thus forming better integrated descriptor. We have used ART neural network algorithm which is unsupervised learning method to achieve dimensionality reduction, thus improving the overall performance of the system. In this work, decision tree algorithms are used for protein classification based on the SCOP (Structural Classification of Proteins) hierarchy. Using the C4.5, with 10-fold-cross-validation, algorithm, separate decision trees are built for each level in the SCOP hierarchy. The SCOP database, was used to evalu ate the classification accuracy of this system.

Important Links:

Go Back