Canonical Classes

CDRs are classified in abYsis using sequence rules which specify loop length and permitted amino acids at required positions. Thus classifications are made for all numbered sequences, whether or not structural data are available.

Summary table

The upper Summary table shows the assigned canonical classes for the query sequence, using the different methods of classification (see below). Mismatches are noted where the match is not exact.

Click the radio button for a particular CDR/method combination.

Detail table

The lower Detail table shows the sequence rules for all the canonical classes corresponding to a particular CDR/method combination. Note that the table only shows classes where the loop length matches that of the query sequence. (See the About / Definitions page for the complete set.)

The title rows of the Detail table show the subset of numbered positions that are relevant for the selected CDR/method combination.

The Query: row shows the residues of the query sequence at these positions and the loop length (AbM definition) used in the classification.

There is then one row for each canonical class of that loop length.

The first column shows the canonical class name. The second column shows the accession for a representative structure in abYsis. The third column shows the required loop length (AbM definition). The following columns show the permitted amino acids at each required position.

Note: For a query sequence to have an exact match, the loop length must agree and the amino acids at each required position must match one of the permitted amino acids. For a similar match, the loop length must agree, but there may be one or more position where the amino acids do not match one of the permitted amino acids. Individual amino acid matches and the length match are highlighted. The entire row is highlighted for the best match, which is either an exact match or the best similar match. Mismatched positions within the best match are indicated in red.


Auto Method

Likely to give the most accurate results because it is based on strict templates derived automatically from available crystal structures. The templates are generated using an automatic protocol involving cluster analysis and analysis of buried hydrophobic and hydrogen bonding residues (Martin and Thornton, J. Mol. Biol. 263(1996),800-815).

The canonical class identifiers consist of the equivalent Chothia class followed by a structural cluster number consisting of the loop length followed by a letter to describe the individual cluster. Where there is no equivalent Chothia class, a '?' appears (e.g. for CDR-H1, there are classes ?/10C and ?/10D).

AbM Method

The AbM key residues are based on Chothia templates (Whitelegg N & Rees AR, Protein Eng. 13(2000),819-24 and Methods Mol. Biol. 248(2004),51-91). The classes are expanded from the Chothia definitions: some additional allowed residues have been specified together with some additional required residues and extra classes. Classes with a * in the name are not defined by Chothia. Modfications and references are detailed for each class.

Strict Method

These classes result from strict application of templates derived from various papers by Chothia et al.


There are some confusing discrepancies in the canonical class numbers used by Chothia. For example CDR-L1 Class 5 is used in earlier papers to refer to a 13-residue lambda light while later papers use class 5 for a 15-residue kappa. Similarly CDR-L1 Class 6 is used in earlier papers to refer to a 14-residues lambda, but later papers use class 6 for a 12-residue kappa. We have used the earlier class numbers.