I am trying to dock the variable domains of an antibody with its protein target in docking 2. The structure of the variable antibody domains are predicted from homology. I have a few different antibody and for all except one this works fine. One however gives the error message Prepack step failed. Check your input files. I have repredicted the structure of the antibody with various programs, but it keep on failing. What can the problem be? Is there any online program (in Rosie for instance) where I can check my input before I try to dock?
There is some diagnostic output available for Rosie runs.
On your runs - even the failed ones - you can click the "[Full set of decoy structures created on this run (FS view)] " link. Then click the "logs" link. In your case you can look at the docking_prepack.errors and docking_prepack.logs files listed there. That should give you more indication of what might be going wrong.
You don't mention a job, so it's hard to figure out which one is yours, but for example, for this job http://rosie.rosettacommons.org/docking2/viewjob/15170 which failed in the prepack stage, the log output http://rosie.rosettacommons.org/fs/15170/output/logs/00000-docking_prepa... indicates "[ERROR] Exception caught by JobDistributor for job proteins_0001chain_id H represents more than one chain!"
The reason for this is that in the middle of chain H there's a "TER" record between residue 99 and 101. This makes Rosetta think that the H chain is actually two chains. In this case, it's not, so if the person who is running this particular run deletes the TER record, it's likely that their error will be fixed. If the H chain *should* be treated as two chains, then they would want to rename one of them to a different letter (so that each chain has a unique letter), and adjust the parameters for the docking run accordingly.