Key for protein coding sequence identification: computer analysis of codon strategy

Abstract
The signal qualifying an AUG or GUG as an initiator in mRNAs processed by E.coli ribosomes is not found to be a systematic, literal homology sequence. In contrast, stability analysis reveals that initiators always occure within nucleic acid domains of low stability, for which a high A/U content is observed. Since no aminoacid selection pressure can be detected at N-terminl of the proteins, the A/U enrichment results from a biased usage of the code degenerascy. A computer analysis Is presented which allows easy detection of the codon strategy. N-terminal codons carry rather systematically A or U in third position, which suggests a mechanism for translation ini tiation and helps to detect protein coding sequences in sequenced DNA.