INDEXING AND MAPPING OF PROTEINS USING A MODIFIED NONLINEAR SAMMON PROJECTION Nonlinear Sammon Projection of Compositional Space of Proteins can Predict Protein Folding Classes

A modified Sammon's algorithm was applied to display a relationship between proteins based on their amino acid composition. In the first stage of the method the 19-dimensional compositional space of representative proteins was mapped into 2-dimensinal space using the original Sammon projection to create a contour map. In the second stage, the contour map was used as a reference for newly projected proteins. Data analysis showed that proteins belonging to the same structural class form characteristic and distinct clusters which can be utilized in prediction of structural classes. However, significant overlapping of the clusters has been observed which may explain the limited success of previous protein folding predictions based solely on amino acid composition. Additionally, the modified Sammon's projections can generate a unique index for each individually projected protein related to its amino acid composition which can be a useful parameter in classification of proteins.