Fig. 2.

(a) Distribution of gene length in the B. subtilis 168 genome. The absence of any overrepresentation of short CDSs supports the view that most if not all gene sequences predicted in the present annotation are authentic. (b) Correspondence analysis of the proteome of B. subtilis. Proteins in the proteome can be separated into two well-identified classes. The green cloud corresponds to proteins that are integral inner-membrane proteins (IIMPs). Note that the IIMP cloud is driven by the opposition between charged amino acids (D, E and K) and hydrophobic ones (F, L, M, W).