HMMSTR-CM database

This database was created for internal use only. It contains the predicted contact maps for 9192 proteins of known structure. The method used to make the predictions is called HMMSTR-CM The color images show the log-likelihood of a contact between any two residues in the sequence using a hot-to-cold color scale. Red positions are likely contacts, blue are unlikely. Clicking on the image shows the true contacts, and a stereo MolScript image. The probabilities are calculated as the log of the number of times a contact was observed between a pair of residues having HMMSTR states p and q respectively, divided by the number of contacts expected between those two states by chance. HMMSTR-CM was coupled with a pathway model to make contact map predictions for CASP5 and CASP6. HMMSTR-CM predictions are carried out automatically by the HMMSTR server, as part of a fully automated ab initio prediction pipeline.

SCOP root

1. all alpha
2. all beta
3. a/b mostly parallel
4. a+b mostly anti-parallel
5. multidomain proteins
6. membrane proteins
7. small proteins
Click to navigate the database using SCOP numbering. Visit SCOP for links to protein specifics.