In an overwhelming number of instances, you will find PDBeFold to be very fast. A typical query - matching a protein of a few hundred residues against the whole PDB or SCOP archive - takes less than a minute, a significant part of which falls on the network communication.
If you do not see the performance we claim, it may be that our server is loaded with other tasks. Please be patient. You can always bookmark the wait (progress) page, and this link will get you to the results when they are ready. It is assured that you get a unique wait page, so that you never mix with other jobs or users.
The performance goes down with the size of your query. It is the total number of Secondary Structure Elements (SSEs) that matters, rather than number of residues or protein volume. You will see only a slight slow down for queries having up to 40 SSEs, bearable for queries of 40 to 60 SSEs, while for larger structures it may take 10 minutes and more.
Keep in mind that PDBeFold can match
- chain-against-chain (default setting)
a particular regime being chosen in the Submission Form. The number of SSEs affecting the performance is that participating in a particular comparison. Thus, matching a protein of 4 chains, 20 SSEs each, chain-against-chain (20 SSEs participating in each comparison) is much faster than structure-against-structure (when all 80 SSEs are treated simultaneously).
The performance also goes down with decreasing the threshold for the minimal number of SSEs to be aligned. This threshold may be specified in the Submission Form, where it is initially set at 70% for both the query and target structures (this means that PDBeFold will look only at hits where at least 70% of the query's SSEs and at least 70% of target's SSEs may be related).
The higher the SSE threshold, the faster PDBeFold performs. Typically, varying the SSE threshold within 70%-100% does not affect the PDBeFold performance significantly, while you will always see a considerable slow down if SSE threshold drops below 30%-40%.
When choosing the SSE threshold, consider that higher values will result in a smaller number of better matches. If you are interested in looking for common substructures of a relatively small size (which pushes you to level the SSE threshold down), consider selecting a domain from the query (available in the Submission Form). PDBeFold will then work only with SSEs left in the domain, which is always faster.