SHREC 2007-Protein Challenge

We propose to use the dataset 27folds, which consists of 633 protein domains divided into 27 folds according to their SCOPid. This dataset was first introduced in [3]. This dataset is chosen on purpose since it is quite difficult for classification.
Each protein domain is specified by the three dimensional coordinates of its atoms.
The three dimensional coordinates are used to compute feature vectors or alignment scores. If feature vectors are computed, a distance measure is used on them to compute the similarity score. Finally, a distance matrix is generated with the distance measure between every two protein domains, which then can be evaluated with the tools from the Princeton Shape Benchmark.
The ground truth is given by the SCOP classification.

Download the TXT file with the name of the SCOP-domains and the SCOP classification here. In this PDF file the SCOP names of the domains are listed.

The PDB files can be downloaded from Brookhavens Protein Database. The specification of PDB files can also be found on this homepage. Please note: We will be only interested in the ATOM part of the PDB files!

Download the domains from the 27folds as a zipped archive containing the domains as PDB files from here. This zip-file was created by Bin Li (lib(at)purdue.edu) from Purdue Univeristy. Thank you Bin Li!! In the zip-file the programs shrec.pl and shrec.id are attached. shrec.pl will read shrec.id, perl and rasmol are needed to run the program, and all original pdb files from RCSB should be put in pdb/ directory. The result will be written into out/ directory.

The SCOP domains of the PDB files are specified according to the dir.des.scop.txt 1.71 file from the SCOP homepage.

In order to visualize protein structures, you can use RasMol or PyMOL.

Here are some example models diplayed as cartoons:

This homepage was created by Maja Temerinac for the SHREC 2007 Protein Challenge.
For more information, please contact:
Maja Temerinac temerina(at)informatik.uni-freiburg.de or
Marco Reisert reisert(at)informatik.uni-freiburg.de