openff-nagl
A framework for learning classical force field parameters using graph convolutional neural networks.
openff-nagl [OPTIONS] COMMAND [ARGS]...
Options
- --n-workers <n_workers>
The number of workers to distribute the labelling across. Use -1 to request one worker per batch.
- Default
1
- --worker-type <worker_type>
The type of worker to distribute the labelling across.
- Default
local
- Options
lsf | local
- --batch-size <batch_size>
The number of molecules to processes at once on a particular worker.
- Default
500
database
CLIs for interacting with databases, such as storing and retrieving molecules.
openff-nagl database [OPTIONS] COMMAND [ARGS]...
retrieve-molecules
Retrieve molecules from database
openff-nagl database retrieve-molecules [OPTIONS]
Options
- --input-file <input_file>
Required The path to the SQLite database (.sqlite) to retrieve the labelled molecules from.
store-molecules
Convert pre-computed molecules to database
openff-nagl database store-molecules [OPTIONS]
Options
- --input-file <input_file>
Required The path to the input molecules. This should either be an SDF or a GZipped SDF file.
label-molecules
Label molecules from SMILES
openff-nagl label-molecules [OPTIONS]
Options
- --input-file <input_file>
Required The path to the input molecules: SDF or smiles. SDFs will be converted to smiles
plot
CLIs for plotting.
openff-nagl plot [OPTIONS] COMMAND [ARGS]...
similarity
openff-nagl plot similarity [OPTIONS]
Options
prepare
CLIs for preparing molecule sets, such as filtering out molecules which are too large or contain unwanted chemistries, removing counter-ions, or enumerating possible tautomers / protomers.
openff-nagl prepare [OPTIONS] COMMAND [ARGS]...
calculate-similarity
Calculate similarity between datasets
openff-nagl prepare calculate-similarity [OPTIONS]
Options
- --output-file <output_file>
Required The path to the SDF file (.sdf) to save the generated conformers in.
- --skip <skip>
Include every `skip`th molecule from each file
- Default
10
generate-conformers
Generate and store conformers
openff-nagl prepare generate-conformers [OPTIONS]
Options
- --input-file <input_file>
Required The path to the input molecules. This should either be an SDF or a GZipped SDF file.
- --output-file <output_file>
Required The path to the SDF file (.sdf) to save the generated conformers in.
partition
Partition molecules into training, validation, test datasets
openff-nagl prepare partition [OPTIONS]
Options
- --training-fraction <training_fraction>
Rough percentage of how much should be in the training set.
- Default
0.7
- --validation-fraction <validation_fraction>
Rough percentage of how much should be in the validation set.
- Default
0.2
- --output-training-file <output_training_file>
Required The path (.sqlite) to save the training set to.
- --output-validation-file <output_validation_file>
Required The path (.sqlite) to save the validation set to.
select
Selects a set of molecules based on the criteria specified by:
[1] Bleiziffer, Patrick, Kay Schaller, and Sereina Riniker. ‘Machine learning of partial charges derived from high-quality quantum-mechanical calculations.’ JCIM 58.3 (2018): 579-590.
openff-nagl prepare select [OPTIONS]
Options
- --n-min-molecules <n_min_molecules>
Minimum number of molecules to select from each atom environment
- Default
4