Add general linear fit target (#131)

* initial cleanup

* remove reporting

* add example

* add upper level imports

* update docs

* fix up general fit target and add docs

* add convenience create_data_module function

* start test

* update examples env and readme

* update tests

* remove unused import

* add apsw pin

* upgrade example ci python

* update notebook

* update changelog

.github/workflows/examples-ci.yaml

@@ -31,7 +31,7 @@ jobs:
  fail-fast: false
  matrix:
  os: [macOS-12, ubuntu-latest]
- python-version: ["3.10", "3.11"]
+ python-version: ["3.11", "3.12"]
  pydantic-version: ["2"]
  include-rdkit: [true]
  include-openeye: [false]

devtools/conda-envs/examples_env.yaml

@@ -22,6 +22,8 @@ dependencies:
  - openff-toolkit >=0.11.1
  - openff-units
  - openff-recharge
+ - openff-qcsubmit
+ - psi4
  - pydantic <3
  - rdkit
 
@@ -37,6 +39,9 @@ dependencies:
  # parallelism
  - dask-jobqueue
 
+ # compatibility
+ - apsw >=3.42
+
 
  # CI
  - nbval

docs/CHANGELOG.md

@@ -19,6 +19,9 @@ The rules for this file:
 ### Authors
 - [@lilyminium]
 
+### Added
+- General linear fit target and example (PR #131)
+
 ### Changed
 - Removed unused, undocumented code paths, and updated docs (PR #132)
 

examples/README.md

@@ -6,3 +6,4 @@ The following examples are available in [the OpenFF NAGL repository](https://git
 
 * [prepare-dataset](https:/openforcefield/openff-nagl/tree/main/examples/prepare-dataset) - Prepare a dataset for training, validating or testing NAGL models from a list of SMILES and the OpenFF Toolkit
 * [train-gnn-notebook](https:/openforcefield/openff-nagl/tree/main/examples/train-gnn-notebook) - Architect, train, and use a simple GCN partial charge model on an alkane test dataset
+* [train-electric-field](https:/openforcefield/openff-nagl/tree/main/examples/train-gnn-to-electric-field) - Prepare a dataset from QM data, set up training and validation sets, create a GNN and train to electric field data

examples/train-gnn-to-electric-field/.gitignore

@@ -0,0 +1,3 @@
+lightning_logs
+tmp.pkl
+api.qcarchive.molssi.org_443

examples/train-gnn-to-electric-field/input-env.yaml

@@ -0,0 +1,16 @@
+name: train-gnn-to-electric-field
+
+channels:
+ - conda-forge
+
+dependencies:
+ - openff-qcsubmit
+ - openff-recharge ==0.5.2
+ - psi4
+
+ - jupyter
+ - tqdm
+ - pip
+
+ - pip:
+ - git+https:/openforcefield/openff-nagl.git@main