Align Module Issues about biosimspace HOT 8 CLOSED

Hi Miroslav,

You'll need to use the feature-mapping branch of Sire to be able to use the latest functionality from BioSimSpace.Align. I've made modifications to the MCS matching algorithm to allow it to return the complete set of matches, rather than the most recent one. There are also modifications to support matching of hydrogen atoms, and matches between hydrogens and heavy atoms. Currently, the most up-to-date branch of BioSimSpace is feature-freenrg, so I'd suggest testing with that.

For your second question: You currently can't write the merged molecule to all formats as the parsers do not know how to interpret it. (The only one is Gromacs.) If you just want to visualise things, say in VMD, you can do the following:

# Assumes you have a merged molecule called "mol".

# Create a property map.
prop_map = {}

# Here we map all molecular properties to those from the lambda = 0 state,
# e.g. "coordinates" --> "coordinates0".
for prop in mol._sire_molecule.propertyKeys():
    if prop[-1] == "0":
        prop_map[prop[:-1]] = prop

# Now save the molecule using the mapped properties.
BSS.IO.saveMolecules("merged", mol, "pdb", property_map=prop_map)

Hope this helps.

from biosimspace.

Hi Lester,

Thanks for the reply. These branches did work out for me in the end (although the alignment is not perfect but better than before). The output to gro / top files is also a very useful functionality!

Now I seem to be having issues with solvate(), which is supposed to neutralise the whole system but still leaves total net charge (I think it is ignoring my perturbed ligand which has the same non-neutral charge for both states). But I think this might be a problem of genion, rather than BioSimSpace. I might try to look into it and I will report back if I find something interesting (solvate() takes a long time for a protein - between 5 and 10 mins so testing is a bit difficult).

Many thanks.

from biosimspace.

Thanks for the update. BioSimSpace only performs an alignment using rigid body translations and rotations. If you want to flexibly align one molecule to the other, you could try using fkcombu.

BioSimSpace should neutralise the system by default. It's possible that there's any issue with the perturbed ligand, or maybe a molecule with a fractional charge (if you didn't use BioSimSpace to parameterise it). This is definitely something that should work, so I'm happy to take a look if you zip up your input files and an example script and post them here.

If you've not already spotted, you can see the intermediate output files of the solvation by passing a work_dir keyword argument to any of the functions, e.g.

solvated = BSS.Solvent.tip3p(molecule=molecule, box=3*[10*BSS.Units.Length.angstrom], work_dir="tmp")

from biosimspace.

Yes, I figured that might be the case, thanks for the suggestion, I might try fkcombu (their server is dead though, I hope it is temporary).

So I wrote a simplified script and tested it again (you never know...) and it still gives a net charge. This time I have redirected all of the output files and I am uploading them as well, because running it takes some time but you are welcome to run the script yourself:
Input.zip

All input files have been prepared outside of BioSimSpace using ambertools and converted with a combination of openbabel and parmed so there shouldn't be any issues with that.

EDIT: When I check the last output from solvate (solvated.top), it looks fine (there is one more chloride ion which neutralises the charge), but if I save the system after that, it removes a chloride ion. Strange...

from biosimspace.

I believe the issue is because the -neutral option doesn't seem to work properly in genion. It's supposed to neutralise the system by adding additional ions above those specified by the -np/-nn and -conc options, but doesn't seem to do this in practice. I've worked around the issue for the case where the user doesn't set a concentration (by just using -np/-nn), but haven't fixed things when they do.

from biosimspace.

If you get a chance, could you possible try this again using the latest commit on feature-mapping. I ran across a similar issue late last week so may have fixed your problem.

from biosimspace.

Sorry, I mean feature-freenrg. Too many feature branches!

from biosimspace.

Sorry for the late reply, I seem to have completely missed my notifications!

I tried it again, it seems to work fine now. Many thanks!

from biosimspace.