Book of Abstracts: Albany 2005
Statistical Assessment of Factors of Curvature Distribution in Prokaryotic Genomes
DNA curvature is known to play a biological role in gene regulation, such as initiation and termination of transcription. Moreover, there are convincing indications that distribution of curved DNA is evolutionarily preserved and mainly determined by temperature selection. We applied the software CURVATURE on 170 completed prokaryotic genomes to predict typical distribution of the DNA curvature around starts and ends of genes. Our main objective was to analyze, in statistically manner, possible correlation between a curvature excess in promoter and terminators regions and genomic and environmental factors. We compiled a database, where we put for every genome its characteristics such as its size in bp, number of genes, A+T content averaged over the complete genome, A+T content averaged over all noncoding sequences and A+T composition of coding sequences. We also placed some characteristics of typical environment: pH, growth temperature and aerobiosis/anaerobiosis. We also put some taxonomic descriptors (Kingdom, Phylum, and Class).
We obtained rather expected results showed a significant influence of growth temperature on curvature distribution upstream to the start of translation. Significant differences in curvature in promoter region were observed between mesophilic and thermophilic or hyperthermophilic genomes but not between thermophilic and hyperthermophilic. Significant correlations between curvature distribution in promoter regions and such genomic characteristics, such as genome size and AT composition, were observed. Cluster analysis, which at the first time was applied to such data, backed by other statistical tests strengthened our earlier results that curved DNA is determined by temperature selection and confirmed our assumption that mesophilic genomes different in the curvature distribution in promoter regions according to their characteristics.
Novel results were obtained regarding possible factors influencing curvature distribution in putative terminator regions. Also, strong positive correlation was found between curvature distributions in promoter and terminator regions.