Figure 3

The under-representation of long mononucleotide repeats in coding regions of M. tuberculosis is a consequence of a context-dependent codon choice. Codons consisting of three identical nucleotides ('homogeneous codons') are avoided at positions followed by one or more nucleotides of the same type. The under-representation increases with increasing number of identical nucleotides following. Each line represents data for one type of nucleotide (indicated by the line color).