Abstract:

Support vector machine is used to predict protein-protein interaction of Homo sapiens through the
protein’s primary sequence. A conjoint triad feature is used to describe amino acids. Experimental results show that
the method can predict PPIs with high accuracy. The predicted results of two classifications of Cys are compared.
The PPI network of human fetal liver is predicted, which supplies important information for experiments.

Key words: classification of amino acids, kernel function, support vector machine, protein-protein interaction