SSP'05 -- Proceedings

	IEEE/SP 13th workshop on Statistical Signal Processing July, 17-20, 2005 - Bordeaux - France

Information regarding the paper

Title

A New Approach for HMM based Protein Sequence Modeling and its Application to Remote Homology Classification

Author(s)

Thomas Plötz	Faculty of Technology, Bielefeld University, Germany
Gernot A. Fink	Faculty of Technology, Bielefeld University, Germany

Get the paper in PDF format

To obtain Acrobat Reader (version 5 minimum required) necessary to his read.

Abstract

In this paper we present a new approach for HMM based protein family modeling using building blocks, namely Sub-Protein Units, estimated by analyzing a feature based sequence representation. Contrary to current Profile HMM based approaches, only the classification relevant parts of protein families are modeled. For a representative task of remote homology classification it could be shown that the number of parameters required for robust models and thus the amount of training data can be decreased substantially. This is especially important for pharmaceutical applications.