Abstract

After extracting more than 780 identified Escherichia coli genes from available data libraries, we investigated the codon usage of the corresponding coding sequences and extended the study of gene classes, thus obtained, to the nature and intensity of short nucleotide sequence selection, related to constraints operating at the nucleotide level. Using Factorial Correspondence Analysis we found that three classes ought to be included in order to match all data now available. The first two classes, as known, encompass genes expressed either continuously at a high level, or at a low level and/or rarely; the third class consists of genes corresponding to surface elements of the cell, genes coming from mobile elements as well as genes resulting in a high fidelity of DNA replication. This suggests that bacterial strains cultivated in the laboratory have been fixed by specific use of antimutator genes that are horizontally exchanged.