Fig 1. Definition of ankyrin arrays.

We searched the whole UniProt database and detected repeats with a structurally-based HMM sequence model. If the detected repeats are separated by less than 67 residues, we defined them as belonging to the same array. In the above example, Sequence A codes for 1 single array, and sequence B codes for two arrays. Finally, we get a Multiple Repeat sequence Alignment (MRA) of more than 1.2 million repeats sequences with exactly 33 positions belonging to specific arrays.