Novel metrics for quantifying bacterial genome composition skews

Bacterial genomeshave characteristic compositional skews, which are differences in nucleotide frequency between the leading and lagging DNA strands across a segment of a genome. It is thought that these strand asymmetries arise as a result of mutational biases and selective constraints, particularly for energy efficiency. Analysis of compositional skewsin a diverse set of bacteria provides a comparative context in which mutational and selective environmental constraints can be studied. These analyses typically require finished and well-annotated genomicsequences. We present three novel metrics for examining genomecomposition skews; all three metrics can be computed for unfinished or partially-annotated genomes. The first two metrics, (dot- skewand cross- skew) depend on sequence and gene annotation of a single genome, while the third metric (residual skew) highlights unusual genomesby subtracting a GC content-based model of a library of genomesequences. We applied these metrics to 7738 available bacterial genomes, including partial drafts, and identified outlier species. A phylogenetically diverseset of these outliers (i.e., Borrelia, Ehrlichia, Kinetoplastibacterium, and Phytoplasma) display similar skewpatterns but share lifestyle characteristics, such as intracellularity and biosynthetic dependence on their hosts. Our novel metrics appear to reflect the effects of biosynthetic constraints and adaptations to life within one or more hosts on genomecomposition. We provide results for each analyzed genome, software and interactive visualizationsat http://db.systemsbiology.net/gestalt/ skew_metrics .