Features
ISPRED4 adopts machine-learning methods (SVM+CRF) to predict interaction state of each residue in the protein surface by extracting several features from protein sequence and structure.
Input
To start using ISPRED4 you simply need to upload a protein 3D structure in PDB format and specify the protein chain you want to analyze.
You can also specificy a RSA threshold to select surface residues (all residues with RSA above the threshold are considered surface residues. Default value is: 0.20).
The server accepts in input a single protein structure.
Output
There are two main sections in the result page:
- Sequence and Structure view:
general information about the protein, including PDB ID, the protein length, the surface lenght.
Residues at the surface or predicted as interafaces are highlighted in blue and red, respectively, both in a sequence view panel and in a interactive 3D-view window. - Detailed report:
On the top-left of the table, there are two buttons to download the results in a text format or PDB format.
Below, there is table with the results for each residue, detailing the residue Accessible Surface Area, the Relative Solvent Accessibility (RSA), the Predicted RSA, depth, protrusion, surface, interface and the computed probability.