Markov Analysis of Protein Sequence Similarities
In our previous work, we explored the use of graph-theoretic spectral methods for clustering protein sequences . The nodes of the graph represent a set of proteins to be clustered into families and/or super-families. Edges between nodes are undirected and weighted by the similarities between proteins. We constructed a novel similarity function based on BLAST scores. The similarity values are in turn used to construct a Markov matrix representing transition probabilities between every pair of connected proteins. By analyzing the perturbations to the stationary distribution of the Markov matrix (as in [6,4]), we partition the graph into clusters. In this paper, we compare our method with TribeMCL, which modifies random walks, by reinforcing strong edges and pruning weak ones, such that clusters emerge naturally from the graph . We compare these two methods with respect to their ease of use and the quality of the resulting clusters.
Unable to display preview. Download preview PDF.
- 1.Altschul, S.F., Gish, W., Miller, W., Myers, E.W., Lipman, D.J.: Basic Local Alignment Search Tool. J. Mol. Bio. 215, 403–410 (1990)Google Scholar
- 4.Meila, M., Shi, J.: A random walks view of spectral segmentation. In: Proc. International Workshop on AI and Statistics (2001)Google Scholar
- 5.Murzin, A.G., Brenner, S.E., Hubbard, T., Chothia, C.: SCOP: a structural classification of proteins database for the investigation of sequences and structures. J. Mol. Biol. 247(4), 536–540 (1995)Google Scholar
- 6.Ng, A., Jordan, M., Weiss, Y.: On Spectral Clustering: analysis and an algorithm. In: NIPS (2001)Google Scholar
- 7.Paccanaro, A., Chennubhotla, C., Casbon, J.: M. Saqi Spectral Clustering of Protein Sequences. In: IJCNN 2003 (2003)Google Scholar
- 10.van Dongen, S.: Graph Clustering by flow simulation. Ph. D. Thesis, University of Utrecht, The NetherlandsGoogle Scholar