The clusters were got based on Non-redundant TR arrays, using
1. For 1-10 bp TRs: the consensus patterns must 100% match, different variants (eg. AAT/ATA/ATT/TAA/TAT/TTA are all variants of AAT) were clustered together.
2. For others: under the parameters of pident ≥ 80 and qcovs ≥ 80, variants and multimers were clustered together.