error during restraint search

[mayank-kohli]

Hello,

I have been getting the following error during restraint search step:

UserWarning: MDAnalysis failed to read: traj=/output/restraint_search/gromacs.xtc, top=/output/restraint_search/gromacs.tpr

_warnings.warn( ╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮ │ /simulation/fep_abfe/sub.py:47 in │ │ │ │ 44 # Run the simulation │ │ 45 restraint_search.start() │ │ 46 # Wait for the simulation to finish (block=True), then analyse the simulation to find th │ │ ❱ 47 restraint = restraint_search.analyse(method='BSS', block=True) │ │ 48 │ │ 49 type(restraint) │ │ 50 restraint.getCorrection() │ │ │ │ /anaconda3/envs/bsstutorials2/lib/python3.10/site-packages/BioSimSpace/Sandpit/Exs │ │ cientia/FreeEnergy/_restraint_search.py:395 in _analyse │ │ │ │ 392 │ │ │ │ 393 │ │ # Return the result of calling the staticmethod, passing in the working │ │ 394 │ │ # directory of this object. │ │ ❱ 395 │ │ return RestraintSearch.analyse( │ │ 396 │ │ │ str(self._work_dir), │ │ 397 │ │ │ self._system, │ │ 398 │ │ │ self._process.getTrajectory(), │ │ │ │ /anaconda3/envs/bsstutorials2/lib/python3.10/site-packages/BioSimSpace/Sandpit/Exs │ │ cientia/FreeEnergy/_restraint_search.py:673 in analyse │ │ │ │ 670 │ │ │ ligand_selection_str += append_to_ligand_selection │ │ 671 │ │ │ │ 672 │ │ if restraint_type.lower() == "boresch": │ │ ❱ 673 │ │ │ return RestraintSearch._boresch_restraint( │ │ 674 │ │ │ │ u, │ │ 675 │ │ │ │ system, │ │ 676 │ │ │ │ temperature, │ │ │ │ anaconda3/envs/bsstutorials2/lib/python3.10/site-packages/BioSimSpace/Sandpit/Exs │ │ cientia/FreeEnergy/_restraint_search.py:794 in _boresch_restraint │ │ │ │ 791 │ │ │ │ ) │ │ 792 │ │ │ │ 793 │ │ elif method == "BSS": │ │ ❱ 794 │ │ │ return RestraintSearch._boresch_restraint_BSS( │ │ 795 │ │ │ │ u, │ │ 796 │ │ │ │ system, │ │ 797 │ │ │ │ temperature, │ │ │ │ anaconda3/envs/bsstutorials2/lib/python3.10/site-packages/BioSimSpace/Sandpit/Exs │ │ cientia/FreeEnergy/_restraint_search.py:1734 in _boresch_restraint_BSS │ │ │ │ 1731 │ │ │ return restraint │ │ 1732 │ │ │ │ 1733 │ │ # Find pairs with lowest SD │ │ ❱ 1734 │ │ pairs_ordered_sd = RestraintSearch._findOrderedPairs( │ │ 1735 │ │ │ u, ligand_selection_str, receptor_selection_str, cutoff │ │ 1736 │ │ ) │ │ 1737 │ │ │ │ /anaconda3/envs/bsstutorials2/lib/python3.10/site-packages/BioSimSpace/Sandpit/Exs │ │ cientia/FreeEnergy/_restraint_search.py:1011 in _findOrderedPairs │ │ │ │ 1008 │ │ │ the trajectory. │ │ 1009 │ │ """ │ │ 1010 │ │ │ │ ❱ 1011 │ │ lig_selection = u.select_atoms(ligand_selection_str) │ │ 1012 │ │ pair_variance_dict = {} │ │ 1013 │ │ │ │ 1014 │ │ # Get all receptor atoms within specified distance of cutoff │ ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯ AttributeError: 'NoneType' object has no attribute 'select_atoms'

Thank you

[lohedges]

Are you able to share the input that triggers the exception? It's impossible for us to debug without more information. (There are unit tests for the restraint search that pass during the CI.) Also, in future, could you please enter the requested information such as the operating system and version numbers for Sire and BioSimSpace.

From the error it appears that MDAnalysis is simply unable to read the trajectory files generated by the GROMACS process, i.e.: /output/restraint_search/gromacs.xtc and /output/restraint_search/gromacs.tpr. Are you able to share these files so that we can investigate what went wrong? Do they load with other trajectory parsers, e.g. MDTraj, or Sire?

[mayank-kohli]

When I used the earlier versions of Sire and BioSimSpace in 2023 for the same protein and ligand complex, I did not encounter this error. However, with the current versions, I am now experiencing this issue.

Version in use: Biosimspace; 2024.2.0 Sire: 2024.2.0 Ubuntu 20.04.6 LTS

gromacs1.zip

please find the files

[lohedges]

Thanks, this is really helpful. Here's the error from MDAnalysis:

In [1]: import MDAnalysis

In [2]: u = MDAnalysis.Universe("gromacs.tpr", "gromacs1.xtc")
---------------------------------------------------------------------------
NotImplementedError                       Traceback (most recent call last)
File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/core/universe.py:110, in _topology_from_file_like(topology_file, topology_format, **kwargs)
    109     with parser(topology_file) as p:
--> 110         topology = p.parse(**kwargs)
    111 except (IOError, OSError) as err:
    112     # There are 2 kinds of errors that might be raised here:
    113     # one because the file isn't present
    114     # or the permissions are bad, second when the parser fails

File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/topology/TPRParser.py:212, in TPRParser.parse(self, tpr_resid_from_one, **kwargs)
    211 try:
--> 212     th = tpr_utils.read_tpxheader(data)                    # tpxheader
    213 except (EOFError, ValueError):

File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/topology/tpr/utils.py:233, in read_tpxheader(data)
    232 fileVersion = data.unpack_int()  # version of tpx file
--> 233 fileVersion_err(fileVersion)
    235 # This is for backward compatibility with development version 77-79 where
    236 # the tag was, mistakenly, placed before the generation.
    237 # These versions are development versions between the 4.5 and 4.6 series.

File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/topology/tpr/utils.py:208, in fileVersion_err(fver)
    207 if fver not in setting.SUPPORTED_VERSIONS:
--> 208     raise NotImplementedError(
    209         f"Your tpx version is {fver}, which this parser does not support, yet "
    210     )

NotImplementedError: Your tpx version is 133, which this parser does not support, yet

During handling of the above exception, another exception occurred:

ValueError                                Traceback (most recent call last)
Cell In[2], line 1
----> 1 u = MDAnalysis.Universe("gromacs.tpr", "gromacs1.xtc")

File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/core/universe.py:356, in Universe.__init__(self, topology, all_coordinates, format, topology_format, transformations, guess_bonds, vdwradii, fudge_factor, lower_bound, in_memory, in_memory_step, *coordinates, **kwargs)
    354 if not isinstance(topology, Topology) and not topology is None:
    355     self.filename = _check_file_like(topology)
--> 356     topology = _topology_from_file_like(self.filename,
    357                                         topology_format=topology_format,
    358                                         **kwargs)
    360 if topology is not None:
    361     self._topology = topology

File ~/.conda/envs/openbiosim/lib/python3.12/site-packages/MDAnalysis/core/universe.py:125, in _topology_from_file_like(topology_file, topology_format, **kwargs)
    121         raise IOError("Failed to load from the topology file {0}"
    122                         " with parser {1}.\n"
    123                         "Error: {2}".format(topology_file, parser, err))
    124 except (ValueError, NotImplementedError) as err:
--> 125     raise ValueError(
    126         "Failed to construct topology from file {0}"
    127         " with parser {1}.\n"
    128         "Error: {2}".format(topology_file, parser, err))
    129 return topology

ValueError: Failed to construct topology from file gromacs.tpr with parser <class 'MDAnalysis.topology.TPRParser.TPRParser'>.
Error: Your tpx version is 133, which this parser does not support, yet

It looks like this is an MDAnalysis issue, i.e. it doesn't support the TPR format from the version of GROMACS that you have installed. Could you try an older version of GROMACS to see if that fixes things. Assuming the file format hasn't changed significantly, we might just be able to patch MDAnalysis to override the version check.

[lohedges]

Just to say that I removed the version check in MDAnalysis and it loads just fine, so that should be an option if you can't get a working environment with an older version of GROMACS.

[mayank-kohli]

Ohh, I have recently updated gromacs to version 2024, along with the biosimspace and sire that explains everything, thank you for the clarification. I ll try it with an older version of gromacs.

[lohedges]

No problem. I'll make a note to expose more exception messages from the trajectory interface since the issue was quite clear once I could see the traceback.