GC3 Stage 1b
Dear GC3 Participant,
We have discovered an error in our preparation of the “example” files we provided on 9/21 for the pose-prediction challenge. The example files are fine for their purpose, but we inadvertently populated them with the protein structures from the actual crystallographic results for the corresponding ligands. As a consequence, if a participant, for some reason, chose to dock into these structures, he or she would have been doing self-docking calculations, rather than cross-docking calculations. We cannot know for sure whether anyone used the example protein structures in their calculations, so this component of the challenge can no longer be viewed as an entirely rigorous comparison of methods. On the other hand, we have no reason to expect anyone would have used these files for docking, and we believe that the results can still be useful and informative, both for the individual submitter and for the larger community. Our thought, therefore, is to go ahead with the normal evaluation of all stage 1 predictions, but to be clear in reporting the results that they can no longer be viewed as rigorously comparable.
So that Grand Challenge 3 will include a pose-prediction challenge whose results can be compared without qualification, we are adding a Stage 1b challenge, in which participants are invited to do self-docking calculations into the protein structures from the true co-crystal structures. This stage is of scientific interest, as it allows docking methodologies to be compared without the complicating factor of protein structure selection.
Stage 1b will open Monday, October 9. The data package for Stage 1b, comprising the protein structures for self-docking, will be available here (https://drugdesigndata.org/about/grand-challenge-3/cathepsin_s) . We propose a deadline of 11:59 pm, Pacific Time, October 23.
We apologize for the mistake and invite you to reply with any thoughts or suggestions.