Related Experiment Videos
Multifractal and correlation analyses of protein sequences from complete genomes
Zu-Guo Yu1, Vo Anh, Ka-Sing Lau
1Program in Statistics and Operations Research, Queensland University of Technology, GPO Box 2434, Brisbane Q4001, Australia. yuzg@hotmail.com
Summary
Protein sequences exhibit complex, multifractal patterns, not random ones. This complexity allows for phylogenetic tree construction using a novel time series analysis method.
Area of Science:
- Bioinformatics
- Computational Biology
- Statistical Physics
Background:
- Previous work introduced measure representations for DNA sequences.
- Protein sequences are analyzed using multifractal dynamics.
Purpose of the Study:
- To introduce new measure representations for protein sequences.
- To analyze the complexity of protein sequences using multifractal analysis.
- To develop a method for phylogenetic tree construction based on protein sequence properties.
Main Methods:
- Introduced two measure representations for protein sequences.
- Performed multifractal analysis on protein sequences from complete genomes.
- Applied correlation analysis to time series representations of protein sequences.
Main Results:
- Protein sequences are not random, exhibiting multifractal characteristics.
- Generalized dimensions (D(q)) spectra and specific heat (C(q)) curves reveal distinct patterns.
- Bacteria show classical phase transition-like C(q) curves, while higher organisms display double-peaked curves.
- Multifractal properties alone are insufficient for classification.
- A novel phylogenetic tree construction method based on correlation analysis proved satisfactory.
Conclusions:
- Protein sequences possess non-random, complex multifractal properties.
- The observed multifractal patterns differ between bacteria and higher organisms.
- A new method for phylogenetic analysis using protein sequence time series is effective.