PRIB '09 Proceedings of the 4th IAPR International Conference on Pattern Recognition in Bioinformatics
| Hi-index | 0.00 |
We introduce a method for comparing protein structures using the notion of residue contexts based on protein Cα-atom backbones. The residue context is derived from the set of vectors from a given Cα-atom to each other Cα-atom in the molecule. A three-dimensional histogram is generated from these vectors, containing a relative distribution of the other Cα-atoms for each Cα-atom on the backbone for a protein. Histograms are compared using the χ2 test, resulting in the cost for matching any two given Cα-atoms in a pair of protein molecules. An optimal alignment is made using the Smith-Waterman algorithm, and a score is calculated based on the length of the alignment and the RMSD, yielding a best alignment that can be displayed in an interactive user interface. Resulting alignments are compared with alignments generated by CTSS, DALI, and CE, yielding different aligned protein regions.