Error running example configs

[jordantwells42]

Hello,

I am getting an error across my own config files and the example config files in the topology and rigidbody modules.

For topoaa I get the following error:

[2022-02-17 13:18:38,065 init INFO] [topoaa] CNS jobs have finished
[2022-02-17 13:18:38,072 libutil ERROR] 100.00% of output was not generated for this module and tolerance was set to 0.00%.
Traceback (most recent call last):
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libutil.py", line 320, in log_error_and_exit
yield
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/clis/cli.py", line 148, in main
workflow.run()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libworkflow.py", line 27, in run
step.execute()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libworkflow.py", line 100, in execute
module.run(**self.config)
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/modules/base_cns_module.py", line 48, in run
self._run()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/modules/topology/topoaa/init.py", line 206, in _run
self.finish_with_error(_msg)
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/modules/init.py", line 186, in finish_with_error
raise RuntimeError(reason)
RuntimeError: 100.00% of output was not generated for this module and tolerance was set to 0.00%.
[2022-02-17 13:18:38,082 libutil ERROR] 100.00% of output was not generated for this module and tolerance was set to 0.00%."

If I set the tolerance to 100% I then get this error, showing that no models were being generated in the topoaa step.

[2022-02-17 13:19:31,112 base_cns_module INFO] Running [rigidbody] module
[2022-02-17 13:19:31,115 init INFO] [rigidbody] crossdock=true
[2022-02-17 13:19:31,115 libutil ERROR] division by zero
Traceback (most recent call last):
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libutil.py", line 320, in log_error_and_exit
yield
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/clis/cli.py", line 148, in main
workflow.run()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libworkflow.py", line 27, in run
step.execute()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/libs/libworkflow.py", line 100, in execute
module.run(**self.config)
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/modules/base_cns_module.py", line 48, in run
self._run()
File "/mnt/c/Users/jorda/desktop/projects/haddock3/src/haddock/modules/sampling/rigidbody/init.py", line 47, in _run
sampling_factor = int(self.params["sampling"] / len(models_to_dock))
ZeroDivisionError: division by zero
[2022-02-17 13:19:31,122 libutil ERROR] division by zero

This occurs both with a local installation and installation in a Docker container. I've included the relevant lines of the Docker container below if that might help debug an error in my installation.

COPY cns_solve_1.3 cns_solve_1.3

RUN git clone --recursive https://github.com/haddocking/haddock3.git &&
cd haddock3 &&
cd src/fcc/src &&
chmod u+x Makefile &&
make &&
cd ../../../ &&
python3.9 -m pip install -r requirements.txt &&
python3.9 setup.py develop --no-deps &&
mkdir -p bin/ && \
ln -s /dependencies/cns_solve_1.3/intel-x86_64bit-linux/source/cns_solve-1111091055.exe bin/cns

[joaomcteixeira]

If the topologies are not getting generated is (likely) because the PDB file has errors. Do you know our pdb-tools python package? Could you run pdb_validate on your input PDBs to see the outcome? I would ask you for your PDB files to inspect them personally, but I believe those are private data, so we respect that.

We are currently working on a PDB pre-processing stage to clean PDBs before running haddock. These are the checks we are implementing. Do you adventure yourself to see manually if you PDB complies with those checks? -> #143

[jordantwells42]

Ah okay, that seems to be the case. I ran pdb_validate.py on e2a-hpr_1GGR.pdb in the examples/docking-protein-protein/data directory and several lines come up with the following type of error:

Line XYZ is short: 76 < 80
'ATOM 3591 1HG GLU B 85 23.217 21.453 23.393 1.00 0.24 B '
^^

I am using windows subsystem for Linux, could this be an issue with the EOL character?

I also tried padding the lines with white space with the following command:
sed -i ':a;/.{80}/!{s/$/ /;ba}' e2a-hpr_1GGR.pdb
which made it pass the pdb_validate check but still fail in the same manner in topoaa

[joaomcteixeira]

Scuse me, are you having problems with the examples folder? All the examples have been running and are running on our site. Are you sure you haven't modified any file by mistake? Are you running the -test.cfg examples? Anything you can provide us will be of great help. Thank you!

[joaomcteixeira]

I am using windows subsystem for Linux, could this be an issue with the EOL character?

In the HADDOCK python shell we use os.linesep as much as possible. Surely there are some \n left but we will correct for that also. I don't know the implications of using a linux subsystem on windows regarding that. We don't support Windows. Can you try with a Virtual Machine? You can allocate as many resources you wish to the virtual machine as well.

Please also note that the main branch is not yet production ready. Are actively working on it right now. Expect changes on the daily/weekly basis.

[jordantwells42]

I am running the examples! Sorry for the miscommunication, I should have been clearer.

I showed the commands I'm using to install Haddock3 into a Docker container in the first post. The installation appears to work fine and CNS also works locally. The only slightly strange thing that occurs is that numba requires numpy<=1.22.2 so I reinstall numpy as 1.22.1 later on.

I've been running the docking-protein-protein-test.cfg recipe from the examples/docking-protein-protein directory using the command: haddock3 docking-protein-protein-test.cfg. I've attached the full log file if that would be helpful.
log.txt

And absolutely! We are aware that things are not production ready, and are ready for any sort of changes that may come with that.

And I will try it in a virtual machine soon and let you know if that changes things! Usually WSL is VM-like enough for things to work but this could very well be the issue.

[amjjbonvin]

Can you manually start your CNS executable? Any error message in the out files in the 00_topoaa directory?

[amjjbonvin]

This is not the problem - the same files / examples run on our side

[jordantwells42]

The cns executable (from haddock3/bin directory) appears to work and brings up the CNSsolve> prompt, as well as typing cns or cns_solve from the cli (once .cns_solve_env_sh has been run).

I checked the 00_topoaa directory and the e2ap_1F3G.out file ends as follows. I'm not sure what the MXRTT parameter is, if its memory based it could be an issue with WSL not being allocated enough RAM?
%AUTOAN error encountered: exceeded MXRTT (RTF) parameter --> recompile program
(CNS is in mode: SET ABORT=NORMal END)

---

ABORT mode will terminate program execution.

---

Program will stop immediately.
============================================================
Maximum dynamic memory allocation: 8201488 bytes
Maximum dynamic memory overhead: 208 bytes
Program started at: 16:14:01 on 17-Feb-2022
Program stopped at: 16:14:01 on 17-Feb-2022
CPU time used: 0.0437 seconds
============================================================

I've also attached the full file here if that helps
e2aP_1F36.txt

Thanks for the help y'all, it is much appreciated.
.

[amjjbonvin]

Ah! You need to recompile CNS. You can download a set of routines to add to the standard version from the following link: https://www.dropbox.com/s/wliubqovuusqdvr/cns.tgz?dl=0 These are distributed with haddock2.4 but we should add them to haddock3 as well or provide clear instructions. Check the README file in that dir. You basically have to copy the files into the main source directory of your cns installation and recompile.

[joaomcteixeira]

Hi @jordantwells42
We hope you have been able to solve the problem after recompiling CNS. I am closing this issue but feel free to reopen it if you have any further questions or if you could not actually solve the problem.
Cheers,

[jordantwells42]

Sorry for the late reply! I just confirmed that it worked after recompilation! Thank you for all of the help. Thank you for closing the issue as well.

[amjjbonvin]

Thanks Jordan We clearly need to provide the modified CNS source code and clear instructions to recompile CNS.