Fig. 5

Robustness of the DI _id procedure. Subsets of alignments were constructed by recurrently removing random groups of sequences from each dataset of repeat pairs. M _eff is the number of effective sequences used in the alignment. a Particular examples of the stability of DI _id assignments as sampling changes on the ANK family. The gray shadow delimits the 1 % fluctuation interval set as a convergence criteria. b Overall stability of the DI _id assignments in several repeat protein families