This dataset is organized hierarchically, first by method, then challenge week, target, and candidate.

Each week folder contains a summary of all RMSD values for the predictions therein (RMSD.csv).

A single prediction for a candidate consists of a pair of PDB and MOL files.

The path for the prediction structures for method METHOD, in year YEAR and week WEEK, for target TARGET_PDBID, candidate category CANDIDATE_CATEGORY, and candidate PDB CANDIDATE_PDBID is:
├── ${METHOD}
│   ├── celpp_week${WEEK}_${YEAR}_dockedresults_${METHOD}
│   │   ├── RMSD.csv
│   │   ├── ${TARGET_PDBID}
│   │   │   ├── ${CANDIDATE_CATEGORY}-${TARGET_PDBID}_${CANDIDATE_PDBID}_docked.mol
│   │   │   ├── ${CANDIDATE_CATEGORY}-${TARGET_PDBID}_${CANDIDATE_PDBID}_docked.pdb

For example:
├── autodockvina
│   ├── celpp_week10_2017_dockedresults_autodockvina
│   │   ├── RMSD.csv
│   │   ├── 1fcz
│   │   │   ├── hiResHolo-1fcz_1fcy_docked.mol
│   │   │   ├── hiResHolo-1fcz_1fcy_docked.pdb


The zipped workflows all require installation of the "d3r" python package via pip. Several further require installation of UCSF Chimera, babel, MGLTools, Schrodinger Suite, or the OpenEye Toolkits. Test files and a test run script are provided within each workflow's test_data directory.