Granular support vector machine based method for prediction of solubility of proteins on overexpression in Escherichia coli

Abstract

We employed a granular support vector Machines(GSVM) for prediction of soluble proteins on over expression in Escherichia coli. Granular computing splits the feature space into a set of subspaces (or information granules) such as classes, subsets, clusters and intervals [14]. By the principle of divide and conquer it decomposes a bigger complex problem into smaller and computationally simpler problems. Each of the granules is then solved independently and all the results are aggregated to form the final solution. For the purpose of granulation association rules was employed. The results indicate that a difficult imbalanced classification problem can be successfully solved by employing GSVM. © Springer-Ver lag Berlin Heidelberg 2007.