You are here

Segfault with docking protocol

3 posts / 0 new
Last post
Segfault with docking protocol


I am trying to run the docking protocol for a high resolution local refinement, but I get a Segmentation fault error. The command is

docking_protocol.linuxgccrelease -partners A_B -docking_local_refine -database $ROSETTA_DB -in:file:l list.txt -ex1 -ex2aro -use_input_sc -out:file:silent decoys_high_res_refinement.silent -out:file:silent_struct_type binary -out:path:score -out:prefix high_res_ref_ -out:nstruct 100

The same happens with the parallalel implementation (*.mpi.*). I ran this before and worked OK.

Attached is the ROSETTA_CRASH.log file. Am I missing something?

Thanks in advance.



Post Situation: 
Mon, 2021-02-08 06:39

Looking at the code I can't see anything off-hand which may be at issue from where you're crashing.

The standard suggestion for debugging SegFaults is to first try running it with a debug build. (So docking_protocol.linuxgccdebug versus docking_protocol.linuxgccrelease.) That has extra checking, which has a chance of giving you a more meaninigful error message.

The other thing I would do - expecially as you said that you've run this before an it's worked - would be to go through your list.txt file and check if it's all inputs in the file which have the issue, or if it's only certain ones. If it's only certain ones, I'd double check how the proteins are laid out, taking particular care to check that the chain letters are as you expect (i.e. there's two chains, one "A" and one "B") and that there's nothing "odd" about the ordering of the chains. (That is, you don't have some odd layout, for example, where you have chain B residues interspersed in the chain A ones.)

I'd also go over the logging output, looking for any warning or error messages. In particular, be aware if Rosetta says it's throwing out a large number of residues from your protein. Depending on what the input PDB is like, you might accidentally throw out all of the residues in one chain. (You can try adding the option `-ignore_zero_occupancy false` to the command line and see if that helps.)

Mon, 2021-02-08 09:25


Thanks for your reply.

As you said, the problem was in the chains. I was working with several pdbs from different sources, with different chains.


Mon, 2021-02-08 09:50