The canonical forms of the CDRs were first introduced for antibodies and later for TCRs by Al-Lazikani et al. (2000). Since then, clusters were also identified by Klausen et al. (2015) using the method proposed by North et al. (2011). Here we use the length-independent clustering method (DBSCAN) by Nowak et al. (2016) to identify the canonical forms, and relate them to previous results. The consensus sequence is based on the amino acid conservation in unique sequences of the cluster, where

• An upper case letter indicates that the amino acid has ≥90% conservation.
• A lower case letter indicates that the amino acid has 20-90% conservation.