Hand out date: March 15, 2000 Due date: March 29, 2000
The NIH Molecules R Us utility for enhanced PDB access (graphical views, etc)
Evaluation measures for Structure Prediction
Construct an example that shows how homology modeling could be applied for structure prediction of a protein sequence. You are encouraged to use the automated comparative modeling server SWISSMODEL. Describe briefly the methods used by this program (or any other method including eyeballing similar structures) to construct the 3-D model of a given protein. Can you give a brief justification for using this method?
Pick one or two sequences from the SWISSPROT database with known secondary structure. Use PSI-PRED to perform secondary structure prediction for the sequences. Use the evaluation measures from CASP3 (i.e. Q3 scores) to evaluate the prediction against actual secondary structure. The actual secondary structure can be obtained from the PDB. If you used any other secondary structure prediction method such as PHD, do you obtain a different result? Explain.
For any target from the CASP3 contest, use any fold recognition method (such as 3D-pssm ) . Identify the fold family and briefly describe what you think the structure might look like given just the fold predictions. Would you need a lot more information before you could make the 3-D structure prediction? If yes, briefly describe the kinds of further analyses that might be required.