The lists were generated using a local version of an algorithm similar
to that employed by pisces.
For SA-search, compared to other protein list generators, we use more
stringent criteria to select the proteins.
In particular we exclude chains missing atoms,
alternate conformations on the backbone, a condition to obtain a
correct encoding of structures using HMM-SA. Chains having breask are split in several independent fragments noted as 5csmA_1, 5csmA_2i for instance.
Proteins determined at a too low resolution are also discarded (more
than 2.5 Angströms).
Since 2006, we do not use separate lists for X-Ray and NMR structures.
You can access the current lists here:
RX&RMN-30
(5986 protein chains or fragments)
RX&RMN-50
(8992 protein chains or fragments)
RX&RMN80
(12010 protein chains or fragments)
Having defined lists of non redundant proteins, we build banks of
HMM-SA encoded structures.
The current lists
were built June 2006.
Note: the culling service is
still under construction.