This page makes the sets of protein structures available for download that were used in a recent assessment of physical scoring functions based on CASP4 predictions (M. Feig, C. L. Brooks: Evaluating CASP4 Predictions with Physical Energy Functions, Proteins (2002), 49, 232-245).

For each CASP4 target the data sets are organized with numbered subdirectories below data that contain the following PDB files for each structure in the set:

casp.pdb	original CASP4 submission
casp-complete.pdb	completed all-atom structure
casp-complete-min.pdb	minimized with restraints for 100 steps
min22cdie.pdb	minimized in vacuum for 300 steps
min22rdie4.pdb	minimized with dist. dep.dielectric for 300 steps
casp-rebuilt.pdb	rebuilt all-atom structure (through lattice)
casp-rebuilt-min.pdb	minimized with restraints for 100 steps
rebuilt-min22cdie.pdb	minimized in vacuum for 300 steps
rebuilt-min22rdie4.pdb	minimized with dist. dep.dielectric for 300 steps

Please note, that only residues that are common to all predictions and where coordinates are available in the native structure were used so that all structures for a given target have the same length and the energy scores can be compared to the native structure.

In addition, for each target there is a file casp4.results that contains the original CASP file names as well as the GDT_TS and SOV scores from the CASP4 website for the structures included in the test sets.

More information on the data sets available in the publication mentioned above.

The data can be downloaded from following the links below either as a single file or separately for each target. Each data file is a tar archive, compressed with gzip.