Computational Biology and Data Mining

This pages contain old supplementary material (June 2009) for the "Secondary structure prediction and threading" lesson in Master in Molecular Medicine; Module IV: Functional Genomics; coordinator Hanspeter Herzel.

Software required

Rasmol: protein structure viewer (Roger Sayle)

• Download from: http://www.rasmol.org/software/RasMol_2.7.4/
• Documentation: http://openrasmol.org/

ClustalX: multiple sequence alignment tool (Des Higgins / Julie Thompson)

• Download software: ftp://ftp.ebi.ac.uk/pub/software/clustalw2
  • windows version: clustalx-2.0.11-win.msi
  • (unix and mac versions also available there)
• Documentation: http://bips.u-strasbg.fr/fr/Documentation/ClustalX/

• Firefox
  • Go to Tools/Options/Content
    • enable Java
    • enable JavaScript
• Explorer
  • Go to Internet options/Security/Custom level/Scripting/Scripting of Java applets (enable)

Verify that you can run BiasViz http://cbdm.mdc-berlin.de/tools/biasviz/

Exercises

1A. Domain analysis of a protein sequence

• Get a protein sequence with a kinase domain using SMART http://smart.embl.de/
  • Does it have other domains?
• Run the sequence in PFAM http://pfam.sanger.ac.uk/ and compare the outputs. 
  • Are PFAM domain annotations different from SMART?
  • Are there any 3D representatives of the domains in your protein?

1B. Structure analysis of a protein sequence

• Get a (small/one domain) protein sequence with known 3D structure from PDB (search Entrez/Structures http://www.ncbi.nlm.nih.gov/sites/entrez?db=Structure)
• Run the sequence in Jpred http://www.compbio.dundee.ac.uk/www-jpred/. Compare to the PDB structure. 
  • Was Jpred accurate?
• Run it in SVMfold. http://www.c2b2.columbia.edu/svm-fold/
  • Is there any predicted fold similar to the PDB structure?

1C. Model of a protein structure

• Find a protein in swissprot (http://www.expasy.org/) with one of these domains: PH, SH2, SH3. Find out its swissprot identifier.
• Search for a model of this protein in MODBase (http://modbase.compbio.ucsf.edu/). If there is no model then go back to previous step and find another protein. Store the model coordinates (save as a.pdb file).
• From MODbase, obtain the coordinates of the template PDB. Save coordinates as b.pdb file.
• Open both files in rasmol and compare the structures. 
  • Where are the main differences?

2A. Transmembrane helix prediction

• Get a protein from SwissProt (http://www.expasy.org/) with transmembrane helices and known localization.
• Run it in TMHMM (http://www.cbs.dtu.dk/services/TMHMM/) and check the output.
  • Is the number of TM helices predicted correct?
  • And the orientation of the protein?

2B. Multiple sequence alignment

• Retrieve at least 8 full length homologues of the transmembrane protein used in the previous exercise using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) (pick them up from different organisms)
• Retrieve their sequences and align them with ClustalX. Save alignment with FASTA format.
• Input alignment to BiasViz (http://cbdm.mdc-berlin.de/tools/biasviz/) and find amino acid biased regions.  
  • Can you see the TMs as biased regions?
  • Of which type?
  • Are there any other biased regions?