Representative datasets in the easy mode are generated using the following constraints:
- Sequence identity* between bound and unbound structures > 97%
- Sequence identity between complexes < 30% 1
- Delete homomultimers (sequence identity between chains < 70%)
- Delete crystal packing complexes and structures in wrong format.
Legend
* Using BLAST to calculate the sequence identity.
1 The sequence identity is used to discard redundant co-crystallized structures.