We introduce the concept of multiresolutions using vector quantization and hidden Markov models as a basis for alignment-free comparison of sequences. Different similarity measures can be discovered at different resolutions of the two sequences. The proposed approach provides a new aspect for studying the complexity of biological data and is effective for real-time processing.
Published in:
Bioinformatics Conference, 2003. CSB 2003. Proceedings of the 2003 IEEE
Date of Conference: 11-14 Aug. 2003