Add DeePMD property tools

[zhaiwenxi]

Summary

• Add deepmd_property_tools for molecular property training and prediction.
• Add pyproject.toml packaging, CLI entry point, and documentation.
• Add pytest-based unit tests and GitHub Actions workflow.

Tests

• python -m pytest tests -v

Summary by CodeRabbit

Release Notes

• New Features
  • Added deepmd-property-tools package for molecular property prediction and training
  • New CLI interface with train and predict subcommands for property workflows
  • Support for CSV and MOL file data formats
  • Configuration-based training with automatic checkpoint management and model evaluation

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Repository UI

Review profile: CHILL

Plan: Pro

Run ID: 4dcc958a-cae8-4bc0-a56c-f8e5268f98ec

📥 Commits

Reviewing files that changed from the base of the PR and between 4e64f8b and e9fe00f.

⛔ Files ignored due to path filters (1)

• deepmd/deepmd_property_tools/DATA/dataset_demo.csv is excluded by !**/*.csv

📒 Files selected for processing (75)

• .github/workflows/property_tools_tests.yml
• deepmd/deepmd_property_tools/DATA/mol_convert/id0.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id1.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id10.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id11.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id12.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id13.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id14.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id15.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id16.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id17.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id18.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id19.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id2.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id20.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id21.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id22.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id23.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id24.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id25.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id26.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id27.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id28.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id29.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id3.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id30.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id31.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id32.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id33.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id34.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id35.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id36.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id37.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id38.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id39.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id4.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id5.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id6.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id7.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id8.mol
• deepmd/deepmd_property_tools/DATA/mol_convert/id9.mol
• deepmd/deepmd_property_tools/DPA3_finetune_hyperparameters.md
• deepmd/deepmd_property_tools/MANIFEST.in
• deepmd/deepmd_property_tools/README.md
• deepmd/deepmd_property_tools/deepmd_property_tools/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/cli.py
• deepmd/deepmd_property_tools/deepmd_property_tools/config/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/config/config_handler.py
• deepmd/deepmd_property_tools/deepmd_property_tools/config/default.json
• deepmd/deepmd_property_tools/deepmd_property_tools/data/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/data/converter.py
• deepmd/deepmd_property_tools/deepmd_property_tools/data/datahub.py
• deepmd/deepmd_property_tools/deepmd_property_tools/data/mol.py
• deepmd/deepmd_property_tools/deepmd_property_tools/models/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/models/property_model.py
• deepmd/deepmd_property_tools/deepmd_property_tools/predict.py
• deepmd/deepmd_property_tools/deepmd_property_tools/predictor.py
• deepmd/deepmd_property_tools/deepmd_property_tools/tasks/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/tasks/trainer.py
• deepmd/deepmd_property_tools/deepmd_property_tools/train.py
• deepmd/deepmd_property_tools/deepmd_property_tools/utils/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/utils/base_logger.py
• deepmd/deepmd_property_tools/deepmd_property_tools/utils/metrics.py
• deepmd/deepmd_property_tools/deepmd_property_tools/utils/util.py
• deepmd/deepmd_property_tools/deepmd_property_tools/weights/__init__.py
• deepmd/deepmd_property_tools/deepmd_property_tools/weights/weighthub.py
• deepmd/deepmd_property_tools/predict_property_20.py
• deepmd/deepmd_property_tools/pyproject.toml
• deepmd/deepmd_property_tools/tests/test_cli.py
• deepmd/deepmd_property_tools/tests/test_config.py
• deepmd/deepmd_property_tools/tests/test_mol.py
• deepmd/deepmd_property_tools/tests/test_predict.py
• deepmd/deepmd_property_tools/tests/test_train.py
• deepmd/deepmd_property_tools/tests/test_trainer.py
• deepmd/deepmd_property_tools/train_property_20.py

---

📝 Walkthrough

Walkthrough

A new deepmd_property_tools package is introduced to DeePMD-kit, providing a Uni-Mol-tools-style interface for molecular property prediction and model training. The package includes configuration management, data pipeline for MOL/CSV ingestion, training/inference orchestration via DataHub, PyTorch model wrappers, CLI entry point, and comprehensive tests with example workflows.

Changes

DeePMD Property Tools Package

Layer / File(s)	Summary
Package metadata and CI configuration pyproject.toml, .github/workflows/property_tools_tests.yml, MANIFEST.in	Build system configuration, GitHub Actions CI for unit tests, and manifest directive to include JSON config files in distribution.
Configuration system with defaults deepmd_property_tools/config/__init__.py, config_handler.py, default.json	JSON-based config management via ConfigHandler with recursive deep-merge support; default DPA3 model, loss, learning-rate schedule, and training runtime parameters.
Data ingestion and preparation pipeline deepmd_property_tools/data/__init__.py, converter.py, mol.py, datahub.py	MOL file parsing with 3D coordinate and element extraction, CSV/MOL dataset loading with atom-overlap filtering, type map derivation, DeepMD mixed-npy preparation with train/valid splitting, and DataHub orchestration for training/inference data flows.
Model loading and prediction system deepmd_property_tools/models/property_model.py, predict.py, predictor.py	PropertyModel wrapper for DeepProperty inference, Predictor for coordinate standardization and padded batch prediction with CSV output, and PropertyPredict high-level wrapper with checkpoint/frozen-model selection logic.
Training orchestration and weight resolution deepmd_property_tools/weights/weighthub.py, tasks/trainer.py, train.py	WeightHub for local path or registry-based pretrained model resolution; Trainer supporting single-process training or distributed torchrun with optional post-training checkpoint freezing; PropertyTrain high-level interface coordinating data preparation, training execution, and config persistence.
Command-line interface deepmd_property_tools/cli.py	argparse-based CLI with train and predict subcommands; handlers wiring parsed arguments to PropertyTrain/PropertyPredict, returning integer exit codes.
Utility modules deepmd_property_tools/utils/base_logger.py, metrics.py, util.py	Shared INFO-level logger with stream handler, regression metrics computation (MAE/RMSE), and directory creation utility.
Documentation README.md, DPA3_finetune_hyperparameters.md	Installation and usage guide with Python and CLI examples; detailed DPA3 finetuning guidance separating must-match model-structure fields from user-customizable training parameters, with example PropertyTrain configuration.
Test suite for configuration and data utilities tests/test_config.py, test_mol.py	ConfigHandler merge behavior with nested dict updates; property parsing, atom-overlap detection, element type ordering, and MOL file reading.
Test suite for inference and training workflows tests/test_predict.py, test_train.py, test_trainer.py, test_cli.py	PropertyPredict checkpoint selection (latest numbered vs frozen model), Trainer torchrun command building, PropertyTrain argument validation and epoch-to-step conversion, and CLI train/predict command dispatching.
Test data molecules and example workflows DATA/mol_convert/id0-39.mol, train_property_20.py, predict_property_20.py	40 RDKit V2000-format MOL structures with 3D coordinates, bonds, and formal charges; end-to-end example scripts demonstrating training with DPA3.2-5M pretraining and subsequent inference.

🎯 3 (Moderate) | ⏱️ ~20 minutes

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Suggested labels

Docs, Examples

Suggested reviewers

• njzjz
• iProzd
• wanghan-iapcm

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}