Prometheus' picture from

Can you improve ProMate?
ProMateus allows you to suggest a new feature that might improve the prediction of the binding site location, and test it on our databases.

Please follow these steps: (For details see ProMateus' help)

1 Choose a database:  
transient-hetero interactions (57 unbound structures, 2.36MB)
protein-DNA interactions (48 bound structures, 2.34MB)
protein-DNA interactions (46 models, 2.71MB)
and it.  

2 Extract your suggested feature on this database.
The file that you upload should be a tar-gzipped collection of files named "new.[pdbId][chain]". Each line in each file should contain a single value, representing the amino-acid or surface dot or atom in the same order as they appear in the database.
If your feature needs training, please use the same partition to 5 folds as appears in the lists files.

3 Describe here your suggested feature:
Database: transient-hetero (unbound) protein-DNA (bound) protein-DNA (models)
value per: surface dot amino acid atom  
value type:  
categorical;   Analyze at the proteins level
continuous;   Use bins.
interface probabilities
find your scoring file  

and, then, click

Optional parameters:
- The bounds for identification of interface patches:  
   The higher between the interface-fracion (0-1) and the absolute bound (0-100)

- an e-mail address to be notified when ProMateus finishes:

To easily try ProMateus download this demo file and use it as input for ProMateus. This demo checks the contribution of the amino acid type to the interface prediction for transient-hetero interactions. (Thus, this is a categorical feature over amino-acids on the transient-hetero DB). The running time is up to 25 minutes, depending on the load of the server. (Since this score has already been considered, it is not supposed to improve the prediction).