Estimate reaction feasibility by difference in energy between energy of reactants and products
timurious/estimator:latest
To build an image:
docker build . -t estimator
change directories in docker-compose file near to "source:" line, absolute path to dir with input.json should be provided. Example of json file is shown in example folder.
speed and precision of CREST calculations, possible options :
-
crest_speed (https://github.com/grimme-lab/crest):
-
- quick
-
- squick
-
- mquick (default)
-
dft:
-
- priroda - use priroda16 for dft calculations (PBEPBE 3z), programm developed by D. Laikov https://rad.chem.msu.ru/~laikov/ru/
-
- pyscf (not finished) - use pyscf package for calculations (B3LYP) None or not mentioned in json (default) - do not perform DFT
-
obabel_fast - speed vs precision in 3D generation by openbabel (http://openbabel.org/wiki/Main_Page)
-
- True - can save some time in generation of 3D by mmff94
-
- False or not mentioned in json (default)
docker compose up
or
docker stack deploy -c docker-compose.yaml estimator
The results will be stored in python .shelve file For each job several types of responses available:
-
ReactionComponents (python named tuple):
-
- index - serial number of reaction
-
- smi - SMILES of reaction
-
- reactants - list of CalcResult (python named tuple) for reactants
-
- products - list of CalcResult (python named tuple) for products
-
- energy_dif - difference in energies between products and reactants (Hartree)
-
- comments
-
- time_s - calculation execution time
-
CalcResult (python named tuple) - result of successful calculations
-
- data - list of strings as lines in XYZ file
-
- min_energy - energy of converged structure
-
- log - full output of calculations, if log option is None or omitted in json
-
FailReport (python named tuple)
-
- initial - initial file
-
- log - output of calculations
-
- step - at which step calculations failed