Minimal working example for `weaver`

[adosar]

Hi,

Thanks for putting this project out! I’m interested in using it, but I haven’t been able to find any usage examples or documentation.

Could you share a minimal working example showing how the module is intended to be used? For example, how to construct a MOF from its building blocks and topology?

That would be really helpful — thanks in advance!

[rochusschmid]

dear antonios,
thanks for your comment. and i have to agree: documentation is unfortunately rather sparse, since it is always lowest on the todo list. so we are happy to get info on what is missing like your post.
in fact in this particular case there is a MWE for building HKUST1 to be found here: doc/weaver/examples/HKUST-1

there you will find the CuPW and btc building blocks (specific mfpx files .. see the special keys in the header) and the tbo topology file tbo.mfpx (see the topo key in the header). These three files can all be downloaded form www.mofplus.org as such ... or other BBs.

the script to run weaver is the gen_hkust.py script. it contains a lot of comments and should hopefully be self explaining.

note, that for a real case of reverse topological approach (RTA) you might detect several identical orientations. this will happen e.g. for MOF-5 but not for HKUST-1. for several orientations we sue a genetic algorithm to search for the global minimum which implies that you have a theoretical method (could be UFF4MOF or something generic, but i do not trust these methods very much, or a system specific FF like MOF-FF or even some GFN1/2 semiempiric method if you can afford it.

please let us know if this worked for you and was helpful. if you want, let us know what MOFs you are intending to weave.

best regards,
rochus

[adosar]

Hi Rochus,

Thank you very much for your detailed and helpful response. I appreciate the pointers and the clarification regarding the available documentation and example files.

I'll take a closer look at the HKUST-1 example and the related scripts, and will let you know if I run into any issues or have further questions. Also, I have to say I really like the name weaver — spot on!

[adosar]

Dear @rochusschmid,

I am trying to build LAMMPS following the instructions described here. However, I don't have available ccmake for the UI configuration. If I run cmake ../cmake & make I get a liblammps.a instead of a liblammps.so. Do you know if it is possible to perform the installation without ccmake?

Thanks in advance!

[rochusschmid]

dear antonios,
the installation instructions on our cmc-tools page on github are focused on our typical "linux desktop" environment in our research group. however, the building of lammps is a complex process depending on what you want to do with it (GPU, what packages etc.) and on what hardware you install. the complete details you can find here.

let me add two more things: installing ccmake is rather simple (either via apt on debian/ubuntu or the like on other distros, or i think you can also get it using conda/mamba from conda-forge). the ccmake is rather handy if you don't want to remember all the -D options on the CLI of cmake.
however, in principle, all can be done with cmake. note, that for pylmps to work (which uses the python integration shipped with lammps) you need to build the shared lib (.so). to compile this you would presumably need something like -D BUILD_SHARED_LIBS, which is one of the switches we set in the ccmake GUI (see the screenshot in the installation instructions).

hope this helps.
regards,
rochus

[adosar]

Thanks for the detailed information @rochusschmid! I manged to successfully install and test LAMMPS using cmake by passing all the necessary -D options.

Now I am trying to build MOF-5 similar to the HKUST-1 example, which runs fine. I have downloaded the pcu.mfpx file from the MOFPlus site and I have converted the building blocks to the required .mfpx format with the x2x2 command. The script is the following (just replacing the names from the gen_hkust.py file):

import weaver

# a weaver.framework instance has to be instanciated with a name
f = weaver.framework('MOF-5')

# first ingredient: Topology. given as topo type mfpx file
f.read_topo('pcu.mfpx')

# second ingredient: building blocks, given as bb type mfpx files
f.assign_bb('0', 'node.mfpx')
f.assign_bb('1', 'edge.mfpx')

# since nets usually come with edge length == 1, we need to scale 
f.autoscale_net(fiddle_factor=1.5) # c-c bond formed 
#f.scale_net(6.0) 

# iterative rotational optimization with 10 trials per vertex
f.scan_orientations(10)
f.write_orientations('MOF-5.orients')

# construct MOF structure from topo+BBs+orientation
f.generate_framework() 
#f.generate_framework(autoscale=True) # uncomment for autoscaling

# write the MOF structure as mfpx file
f.framework.write('MOF-5.mfpx')
# write the MOF structure as tinker xyz file
f.framework.write('MOF-5-.txyz')

However, when running the script I get the following error:

ImportError: Impossible to load ASE
You are molsys commit (short): c1fdf5d
Found uncommited changes: commit hash is not accurate!
Traceback (most recent call last):
  File "/home/asarikas/temp/test_weaver/example.py", line 10, in <module>
    f.assign_bb('0', 'node.mfpx')
  File "/home/asarikas/sandbox/cmc-tools/src/weaver/framework.py", line 238, in assign_bb
    self.bbs[vertex].read(fname,ftype=ftype)
  File "/home/asarikas/sandbox/cmc-tools/src/molsys/mol.py", line 234, in read
    formats.read[ftype](self,f,**kwargs)
  File "/home/asarikas/sandbox/cmc-tools/src/molsys/fileIO/mfpx.py", line 163, in read
    connector, connector_atoms, connector_types = parse_connstring(mol,con_info)
                                                                       ^^^^^^^^
UnboundLocalError: cannot access local variable 'con_info' where it is not associated with a value

Do you know what might causing this error? Below are the .mfpx files:

# node.mfpx
# type xyz
5
  1 zn     1.099920   -0.090600   -0.017060   zn                      -1                  -1       2  
  2 o      2.918730   -0.090610   -0.017050   o                       -1                  -1       1       3       4       5  
  3 zn     3.525000    0.263140   -1.694960   zn                      -1                  -1       2  
  4 zn     3.525000    1.185630    1.128260   zn                      -1                  -1       2  
  5 zn     3.524990   -1.720590    0.515540   zn                      -1                  -1       2  

# edge.mfpx
# type xyz
16
  1 o      1.368010    0.074500   -0.614270   o                       -1                  -1       2  
  2 c      2.420530   -0.035300    0.096450   c                       -1                  -1       1       3       4  
  3 o      2.859340    0.861040    0.889800   o                       -1                  -1       2  
  4 c      3.210110   -1.328010   -0.009920   c                       -1                  -1       2       5       9  
  5 c      2.795300   -2.364970   -0.861130   c                       -1                  -1       4       6      13  
  6 c      3.525890   -3.558640   -0.962270   c                       -1                  -1       5       7      14  
  7 c      4.695310   -3.761000   -0.211110   c                       -1                  -1       6       8      10  
  8 c      5.109320   -2.725050    0.641420   c                       -1                  -1       7       9      15  
  9 c      4.381110   -1.529420    0.738740   c                       -1                  -1       4       8      16  
 10 c      5.482560   -5.055060   -0.317850   c                       -1                  -1       7      11      12  
 11 o      6.533140   -5.168110    0.395860   o                       -1                  -1      10  
 12 o      5.046330   -5.949160   -1.115040   o                       -1                  -1      10  
 13 h      1.891000   -2.237310   -1.452280   h                       -1                  -1       5  
 14 h      3.183340   -4.342800   -1.633350   h                       -1                  -1       6  
 15 h      6.011790   -2.853970    1.235250   h                       -1                  -1       8  
 16 h      4.726890   -0.742320    1.405210   h                       -1                  -1       9  

[rochusschmid]

dear antonios,

as you can see from the HKUST1 example, the mfpx files for the BBs are special. they need some additional keywords to define the connecting atoms and the center of the BB.

for example, the CuPW.mfpx in the example has the following header:

# type xyz
# bbcenter com
# bbconn 4*4 7*7 10*10 13*13 

indicating that the center of the BB is the center of mass and the connector atoms are 4, 7, 10, and 13. you will see that they also have special atomtypes with a wildcard indicating that one next neighbor is undefined.

next, you need to recognize that you can (normally) truncate into BBs only through a single bond. so for MOF-5 you can not use Zn4O as a BB but you must use Zn4O(CO2)6 and a C6H4 as a 2cn linker for the edge.

note, that these BBs can be downloaded from MOFplus.org. (for the inorganic node it is here).

the most important problem to realize is that MOF-5 is ill-defined by the topology pcu. please have a look at our old work on this topic: Conformational Isomerism in the Isoreticular Metal Organic Framework Family: A Force Field Investigation. you will find that there are 2 possible orientations of the inorganic BB to be inserted (Td is missing the center of inversion of Oh). for weaving MOF-5 you need either to manually specify the orientation or do a full blown RTA.

in addition to that, you need to add another vertex for the edge. you read a file pcu.mfpx. did you get that from MOFplus.org? pcu has only one vertex. so you will need to add another linker vertex.
you can use f.add_linker_vertex('1') to add a vertex with name 1 on all edges and then assign the linker BB to that edge.

so all in all you should be able to easily generate one random isoreticular isomer of MOF-5 but likely not the most stable one with the tetrahedra inserted in alternating directions.

i will try to make an example how to do an RTA with weaver for MOF-5. (might take a few days, cause i am busy right now)

cheers,
rochus