Search JCSG  
Search JCSG          
       SCWRL Server       |       About

>>About SCWRL Server

      One of the essential steps of homology modeling is replacing protein residues of a template with the residues of a target sequence according to their mutual alignment. Further steps such as building loops, and optimization of backbone geometry for the entire model are known to be less reliable than the first one. So, for many purposes, including the molecular replacement phasing method of protein crystallography, the simplest models obtained by side-chain replacement are also the most useful.

      Replacing side-chains requires non-trivial optimization of side-chain packing. SCWRL, written by Roland Dunbrack, Adrian Canutescu, Mike Bower et al. (Ref. 1), is currently the most advanced algorithm for predicting sidechain conformation.

      At JCSG we developed an automated web server that uses the SCWRL program, which builds a protein model based on a provided alignment and homologous protein structure. The server also automatically addresses small discrepancies between the sequence derived from the template structure and full template sequence (resulting from incomplete structure solution or mutations) by aligning these sequences with BLAST (Ref. 2) and introducing appropriate changes in the target-template alignment.

      Currently, the FFAS03 profile-profile alignment server located at http://ffas.ljcrf.edu is connected to the SCWRL server. After completion of the homology search at the FFAS server a user can automatically start model building with SCWRL server.


To use the SCWRL modeling server the user should provide: 1) a template structure in the .pdb format, 2) an alignment of the target sequence with the template sequence.

The Template File Structure

There are two ways to provide the template structure:
  • Uploading a .pdb file (with only one chain)
  • Providing a PDB code and a chain symbol for automated retrieval from the PDB database
  • (Do not attempt to use both ways at once.)

    The Alignment

    The alignment consists 3of a target sequence to be superimposed on a template backbone, and the sequence of the template backbone. Please place the target sequence on top and the template sequence underneath.

    The Output

    The server output includes the alignment between the full template sequence and the template sequence corresponding to atom entires in the template structure. This alignment shows which regions of the template structure are not known (the final model may have additional gaps there). Before model building the original target-template alignment is automatically adjusted to take ther discrepancies into account. Subsequently two models are built with the SCWRL program. The output from the SCWRL program is displayed.

    The All Atom Model

    The all atom model is built by replacing all template residues with target resides according to the target-template alignment.

    Mixed Model

    The mixed model is built by replacing with serine all template residues which are not conserved between the target and the template except for glycines and alanines. Mixed models are used in the molecular replacement phasing method of crystallography, when side-chain conformations in all atom models are not accurate enough for that purpose.

    A. A. Canutescu, A. A. Shelenkov, and R. L. Dunbrack, Jr. A graph theory algorithm for protein side-chain prediction. Protein Science 12, 2001-2014 (2003).
    S.F. Altschul et al., "Basic local alignment search tool," Journal of Molecular Biology, 215:403-10, 1990.

    Calvin Chen, Lukasz Jaroszewski
    Joint Center for Structrual Genomics
    @ Center for Research in Biological Systems (UCSD)

    JCSG Menu