Linear regressions of the number of protein families per cell location in function of the average number of genes in a clade.

<p>Data points were removed for clarity. The absolute, but not the relative, frequency of proteins in each localization increases with the number of genes in genomes. Only the percentage of periplasmic proteins (indicated with an *) shows a significant correlation with genome size (Spearman’s rho p<0.001, after Bonferroni correction for multiple tests). The slope of the increase in the number of proteins with a given localization was not significantly different from the average trend for the other cell localizations (p>0.05, same test). Abbreviations of cell localizations: cytoplasm (Cyt), inner membrane (IM), periplasm (Per, Proteobacteria), cell wall (CW, Firmicutes), outer membrane (OM, Proteobacteria) and extracellular (Extr).