Protein Structure Visualization by Dimension Reduction and Texture Mapping

Among the biological macromolecules, proteins have attracted special attention from the scientific community due to their rich functional roles. The ability to visualize and manipulate macromolecular structures on graphical display devices has facilitated the identification and analysis of these macromolecules. Structural analyses of the proteins often provide important insights into their biochemical functions. However, such analysis is often limited by the representation of protein structures and the corresponding computational resource requirements. In this study, we focus on the molecular surface of the proteins and investigate computationally and visually effective representations to serve a number of visualization and analysis purposes. Specifically, we "unfold" the protein surface onto a planar space, while preserving the local surface features as much as possible. In contrast to classical cartographic projections, our approach is able to preserve local shape features. Several biochemical properties associated with each surface point are mapped to generate a two dimensional map of these features. The 3D-2D mapping of the surface vertices has also been utilized to texture-map an arbitrary image back onto the protein structure to facilitate the visualization of the 3D structure.