Analysis of protein structure topologies by graph theoretical methods

Similarity between proteins can be defined in many different ways, including occurrence in closely related organisms, interaction with the same ligands, similar amino acid sequences or 3D structures. In this study the Protein Topology Graph Library (PTGL), an existing database which allows proteins to be compared on the super-secondary structure level, is extended to support ligand information. The PTGL uses graphs to model protein topologies. Their vertices represent the secondary structure elements (SSEs) α-helix and β-sheet while the edges model contacts and spatial relations between these SSEs. Similarity between proteins is defined via common substructures in their protein graphs. A new vertex type for ligands is added to the protein model in this work. Substantial parts of the existing programs had to be changed to compute ligand contacts on the different levels or protein structure. Programs that allow for automated updating of the PTGL have also been implemented. Furthermore, the new desktop application Visualization of Protein Graphs (VPG), which computes protein graphs from PDB files and the SSE assignments of the DSSP algorithm, is presented. The protein graphs can also be visualized and written to image files. The program supports the computation of ligand contacts and allows the results to be saved to a database or to text files in a custom format for protein graphs. Optionally, coiled regions in proteins can also be included in the graph representations. Both the new version of the PTGL and the VPG software package enable comparison of proteins on the super-secondary structure level including ligands and are thus of interest in the fields of proteomics, drug design and molecular medicine.

Example_protein_graph