Numbering schemes
IMGT: has 128 possible positions for *all* antigen receptor types. These are supposed to be structurally equivalent. In theory these are supposed to account for all possible positions. However, insertions are possible especially at CDR3. ANARCI gives the insertion codes as letters. Insertions at CDR3 are placed symmetrically about imgt positions 111 and 112. e.g. 111-ABCD DCBA-112.
Kabat: is defined for heavy and light chain antibody chains only (in ANARCI). Positions in the two chain types are not equivalent. Insertions occur at specific positions and can occur in both the framework and the CDRs. They are annotated from A->Z. e.g 100ABCDEFGH 101.
Chothia: is defined for heavy and light chain antibody chains only (in ANARCI). Numbering in the two chain types are not equivalent. Insertions occur at specific positions and can occur in both the framework and the CDRs. They are annotated from A->Z. e.g 100ABCDEFGH 101. It differs to the Kabat scheme by the position it places the insertions at CDRH1.
Martin: is defined for heavy and light chain antibody chains only. Numbering in the two chain types are not equivalent. Insertions occur at specific positions and can occur in both the framework and the CDRs. They are annotated from A->Z. e.g 100ABCDEFGH 101. It differs to the Chothia scheme by the position it places the certain insertions in the framework. It is also referred to as the enhanced Chothia scheme.
AHo: has 149 possible for *all* antigen receptor types. These are supposed to be structurally equivalent. The AHo scheme's large number of positions is supposed to account for all possible positions without the need for specifying insertion positions. In ANARCI, extra residues in the framework may be represented by insertions although these are unlikely to occur in natural sequences.
Wolfguy: is defined for heavy and light antibody chains. Numbering in the two chain types are not equivalent. Different regions of the domain are denoted by a range of numbers. Many possible positions in the CDRs mean that insertion codes are not required. In ANARCI, extra residues in the framework may be represented by insertions although these are unlikely to occur in natural sequences. The CDRs are numbered in an 'up' and 'down' direction. The annotations of CDRL1 is defined according to the canonical structure. In ANARCI this is recognised by taking a sequence similarity to hard coded sequence motifs for different lengths.
Reference
Dunbar J, Deane CM. ANARCI: antigen receptor numbering and receptor classification. Bioinformatics. 2016;32(2):298-300