• Archaea
• Bacteria
• Eukaryota
• Viruses

• AA
• CBM
• CE
• GH
• GT
• PL

• CAZyme Classifications

• AA
• CBM
• CE
• GH
• GT
• PL

• AA
• GH
• PL

Welcome to CAZyme3D

CAZyme3D, as the first CAZyme three-dimensional (3D) structure database, has the following unique features:

1. CAZyme3D contains 870,740 CAZymes (from CAZyDB) and their 3D structures (96% from AlphaFoldDB and the rest are self-predicted using AlphaFold2), forming the Whole CAZyme3D database.


2. CAZyme3D has a subset of CAZymes 3D structures (the CAZyme3D ID50 database) from 188,574 CAZyme representative sequences as a non-redundant CAZyDB (CD-HIT identity threshold 0.5 to generate sequence clusters CAZy ID50).


3. Keyword search and display options in protein structure page are provided for CAZyme families, CAZy ID50 representatives, GenBank ID, UniProt ID, predicted structure pLDDT score, and its visualization (including predicted carbohydrate-binding sites and dbCAN domains highlighted in the 3D structure), which can all be batch downloaded.


4. CAZyme3D allows users submit their own protein seqs for TM-Vec structure–aware similarity search against CAZyme3D sequences (seq vs seq), or submit PDB structures (solved or AlphaFold2 predicted) for Foldseek structure–structure similarity search against CAZyme3D 3D structures (pdb vs pdb). Both searches will allow more sensitive remote CAZyme homology searh beyond the traditional sequence similairty (e.g., BLAST and HMMER).


5. For the 188,574 CAZyme3D ID50 structures, we also generated intra-family level structural similarity and visualization (e.g., all ID50 structures of GH5 are subject to an all vs all Foldseek clustering, total 25,757 structural clusters for all ID50 structures). Intra-family structural similarity will help understand how sequence clusters (like seq subfamilies, e.g., GH5_1 and GH5_2) of each family (GH5) is related to each other at the 3D structure level.


6. We also performed the inter-family structural similarity comparisons and clustering for the 461 families using ID50 structures of each family. Specifically, the structural similarities between each pair of CAZyme families within the same class were calculated. For example, the GH class has 184 CAZyme families. We calculated the average TM-score for each pair of GH families (e.g., GH5 vs. GH13) and plotted the hierarchical clustering of all 184 GH families as a heatmap based on their pair-wise TM-scores.

Start exploring the fascinating world of CAZymes with CAZyme3D.

Page view statistics