ATOMScience-org / AMPL

The ATOM Modeling PipeLine (AMPL) is an open-source, modular, extensible software pipeline for building and sharing models to advance in silico drug discovery.
MIT License
131 stars 59 forks source link
cheminformatics chemistry chemistry-discovery machine-learning math-physics nci-doe-collaboration-capability neural-network python

License LinkedIn

| Install | Docker | Tutorials | Features | Pipeline parameters | Docs |

ATOM Modeling PipeLine (AMPL) for Drug Discovery

An open-source, end-to-end software pipeline for data curation, model building, and molecular property prediction to advance in silico drug discovery.

Created by the Accelerating Therapeutics for Opportunities in Medicine (ATOM) Consortium

Static BadgeStatic BadgeStatic BadgeStatic BadgeStatic BadgeStatic Badge

Check out our new tutorial series that walks through AMPL's end-to-end modeling pipeline to build a machine learning model! View them in our docs or as Jupyter notebooks in our repo.

The ATOM Modeling PipeLine (AMPL) extends the functionality of DeepChem and supports an array of machine learning and molecular featurization tools to predict key potency, safety and pharmacokinetic-relevant parameters. AMPL has been benchmarked on a large collection of pharmaceutical datasets covering a wide range of parameters. This is a living software project with active development. Check back for continued updates. Feedback is welcomed and appreciated, and the project is open to contributions! An article describing the AMPL project was published in JCIM. For those without access to JCIM, a preprint of the article is available on ArXiv. Documentation is available here.


Table of contents

Useful links

Create pip environment

1. Create a virtual env with Python 3.9

Make sure to create your virtual env in a convenient directory that has at least 12Gb space.

Go to the directory where the new environment directory be installed in. Define an environment variable - "ENVROOT".

export ENVROOT=~/workspace # for LLNL LC users, use your workspace
or
export ENVROOT=~ # or the directory as your environment root
cd $ENVROOT

We use "workspace" and "atomsci-env" as an example here.

# LLNL only: 
# module load python/3.9.12
cd $ENVROOT
python3.9 -m venv atomsci-env

2. Activate the environment

source $ENVROOT/atomsci-env/bin/activate

3. Update pip

pip install pip --upgrade

4. Clone AMPL repository

git clone https://github.com/ATOMScience-org/AMPL.git 

5. Install pip requirements

Depending on system performance, creating the environment can take some time.

Note: Based on which environment (CPU or CUDA) to run on, only run one of the following:

First load the CUDA module. Then run cuda specific package install.

cd AMPL/pip
# LLNL only: 
# module load cuda/11.8
pip install -r cuda_requirements.txt

If you get out of memory errors, try setting these environment variables:

export LD_LIBRARY_PATH=<your_env>/lib:$LD_LIBRARY_PATH
export PYTHONUSERBASE=<your_env>
export OPENBLAS_NUM_THREADS=1
export OMP_NUM_THREADS=48
export PYTORCH_HIP_ALLOC_CONF=gargage_collection_threshold:0.9,max_split_size_mb:128
export TF_FORCE_GPU_ALLOW_GROWTH=true

6. (Optional) LLNL LC only: if you use model_tracker, install atomsci.clients

# LLNL only: required for ATOM model_tracker
pip install -r clients_requirements.txt

Install AMPL

Run the following to build the "atomsci" modules. This is required.

# return to AMPL parent directory
cd ..
./build.sh
pip install -e .

Installation Quick Summary

export ENVROOT=~/workspace           # set ENVROOT example
cd $ENVROOT                          # go to a convenient home directory
# LLNL only:
# module load python/3.9.12

python3.9 -m venv atomsci-env        # create environment with Python 3.9
source $ENVROOT/atomsci-env/bin/activate 
pip install pip --upgrade               

git clone https://github.com/ATOMScience-org/AMPL.git # clone AMPL
cd AMPL/pip    
# LLNL only:
# If use CUDA: 
# module load cuda/11.8                         
pip install -r cpu_requirements.txt    # install cpu_requirements.txt OR cuda_requirements.txt  

# LLNL only: required for ATOM model_tracker
# pip install -r clients_requirements.txt

cd ..                                   
./build.sh                            
pip install -e .                        

Create jupyter notebook kernel (optional)

To run AMPL from Jupyter Notebook, with your environment activated. To setup a new kernel:

python -m ipykernel install --user --name atomsci-env

Install with Docker

Note: Be sure to save any work you want to be permanent in your workspace folder. If the container is shut down, you'll lose anything not in that folder.

Uninstall

To remove AMPL from a pip environment use:

pip uninstall atomsci-ampl

To remove an entire virtual environment named "atomsci-env":

rm -rf $ENVROOT/atomsci-env

To remove cached packages and clear space:

pip cache purge

AMPL Features

AMPL enables tasks for modeling and prediction from data ingestion to data analysis and can be broken down into the following stages: ### 1. Data curation - Generation of RDKit molecular SMILES structures - Processing of qualified or censored data processing - Curation of activity and property values ### 2. Featurization - Extended connectivity fingerprints (ECFP) - Graph convolution latent vectors from DeepChem - Chemical descriptors from Mordred package - Descriptors generated by MOE (requires MOE license) ### 3. Model training and tuning - Test set selection - Cross-validation - Uncertainty quantification - Hyperparameter optimization ### 4. Supported models - scikit-learn random forest models - XGBoost models - Fully connected neural networks - Graph convolution models ### 5. Visualization and analysis - Visualization and analysis tools

Details of running specific features are within the parameter (options) documentation. More detailed documentation is in the library documentation.


Running AMPL

AMPL can be run from the command line or by importing into Python scripts and Jupyter notebooks.

Python scripts and Jupyter notebooks

AMPL can be used to fit and predict molecular activities and properties by importing the appropriate modules. See the examples for more descriptions on how to fit and make predictions using AMPL.

Pipeline parameters

AMPL includes many parameters to run various model fitting and prediction tasks.

Library documentation

AMPL includes detailed docstrings and comments to explain the modules. Full HTML documentation of the Python library is available with the package at https://ampl.readthedocs.io/en/latest/.

More information on AMPL usage


Tests

AMPL includes a suite of software tests. This section explains how to run a very simple test that is fast to run. The Python test fits a random forest model using Mordred descriptors on a set of compounds from Delaney, et al with solubility data. A molecular scaffold-based split is used to create the training and test sets. In addition, an external holdout set is used to demonstrate how to make predictions on new compounds.

To run the Delaney Python script that curates a dataset, fits a model, and makes predictions, run the following commands:

source $ENVROOT/atomsci-env/bin/activate # activate your pip environment.
cd atomsci/ddm/test/integrative/delaney_RF
pytest

Note: This test generally takes a few minutes on a modern system

The important files for this test are listed below:

More example and test information


Advanced AMPL usage

Command line

AMPL can fit models from the command line with:

python model_pipeline.py --config_file filename.json # [filename].json is the name of the config file

To get more info on an AMPL config file, please refer to:

Hyperparameter optimization

Hyperparameter optimization for AMPL model fitting is available to run on SLURM clusters or with [HyperOpt](https://hyperopt.github.io/hyperopt/) (Bayesian Optimization). To run Bayesian Optimization, the following steps can be followed. 1. (Optional) Install HyperOpt with "pip install hyperopt" 2. Pre-split your dataset with computed_descriptors if you want to use Mordred/MOE/RDKit descriptors. 3. In the config JSON file, set the following parameters. - "hyperparam": "True" - "search_type": "hyperopt" - "descriptor_type": "mordred_filtered,rdkit_raw" (use comma to separate multiple values) - "model_type": "RF|20" (the number after | is the number of evaluations of Bayesian Optimization) - "featurizer": "ecfp,computed_descriptors" (use comma if you want to try multiple featurizers, note the RF and graphconv are not compatible) - "result_dir": "/path/to/save/the/final/results,/temp/path/to/save/models/during/optimization" (Two paths separated by a comma) RF model specific parameters: - "rfe": "uniformint|8,512", (RF number of estimators) - "rfd": "uniformint|8,512", (RF max depth of the decision tree) - "rff": "uniformint|8,200", (RF max number of features) Use the following schemes to define the searching domains method|parameter1,parameter2... method: supported searching schemes in HyperOpt include: choice, uniform, loguniform, uniformint, see https://github.com/hyperopt/hyperopt/wiki/FMin for details. parameters: - choice: all values to search from, separated by comma, e.g. choice|0.0001,0.0005,0.0002,0.001 - uniform: low and high bound of the interval to serach, e.g. uniform|0.00001,0.001 - loguniform: low and high bound (in natural log) of the interval to serach, e.g. uniform|-13.8,-6.9 - uniformint: low and high bound of the interval as integers, e.g. uniforming|8,256 NN model specific parameters: - "lr": "loguniform|-13.8,-6.9", (learning rate) - "ls": "uniformint|3|8,512", (layer_sizes) - The number between two bars (|) is the number of layers, namely 3 layers, each one with 8~512 nodes - Note that the number of layers (number between two |) can not be changed during optimization, if you want to try different number of layers, just run several optimizations. - "dp": "uniform|3|0,0.4", (dropouts) - 3 layers, each one has a dropout range from 0 to 0.4 - Note that the number of layers (number between two |) can not be changed during optimization, if you want to try different number of layers, just run several optimizations. XGBoost model specific parameters: - "xgbg": "uniform|0,0.4", (xgb_gamma, Minimum loss reduction required to make a further partition on a leaf node of the tree) - "xgbl": "loguniform|-6.9,-2.3", (xgb_learning_rate, Boosting learning rate (xgboost's "eta")) 4. Run hyperparameter search in batch mode or submit a slurm job. ``` python hyperparam_search_wrapper.py --config_file filename.json ``` 5. Save a checkpoint to continue it later. To save a checkpoint file of the hyperparameter search job, you want to set the following two parameters. - "hp_checkpoint_save": "/path/to/the/checkpoint/file.pkl" - "hp_checkpoint_load": "/path/to/the/checkpoint/file.pkl" If the "hp_checkpoint_load" is provided, the hyperparameter search will continue from the checkpoint.

Advanced testing

Running all tests

To run the full set of tests, use Pytest from the test directory:

source $ENVROOT/atomsci-env/bin/activate # activate your pip environment. "atomsci" is an example here.
cd atomsci/ddm/test
pytest

Running SLURM tests

Several of the tests take some time to fit. These tests can be submitted to a SLURM cluster as a batch job. Example general SLURM submit scripts are included as `pytest_slurm.sh`. ```bash source $ENVROOT/atomsci-env/bin/activate # activate your pip environment. "atomsci-env" is an example here. cd atomsci/ddm/test/integrative/delaney_NN sbatch pytest_slurm.sh cd ../../../.. cd atomsci/ddm/test/integrative/wenzel_NN sbatch pytest_slurm.sh ```

Running tests without internet access

AMPL works without internet access. Curation, fitting, and prediction do not require internet access. However, the public datasets used in tests and examples are not included in the repo due to licensing concerns. These are automatically downloaded when the tests are run. If a system does not have internet access, the datasets will need to be downloaded before running the tests and examples. From a system with internet access, run the following shell script to download the public datasets. Then, copy the AMPL directory to the offline system. ``` cd atomsci/ddm/test bash download_datset.sh cd ../../.. # Copy AMPL directory to offline system ```

AMPL tutorials

Please follow link, "atomsci/ddm/examples/tutorials", to access a collection of AMPL tutorial notebooks. The tutorial notebooks give an exhaustive coverage of AMPL features. The AMPL team has prepared the tutorials to help beginners understand the basics to advanced AMPL features, and a reference for advanced AMPL users.


Development

Installing the AMPL for development

Using "pip install -e ." will create a namespace package in your environment directory that points back to your git working directory, so every time you reimport a module you'll be in sync with your working code. Since site-packages is already in your sys.path, you won't have to fuss with PYTHONPATH or setting sys.path in your notebooks.

Code Push Policy

It's recommended to use a development branch to do the work. After each release, there will be a branch opened for development.

The policy is

  1. Create a branch based off a development ("1.6.0 "for example) or "master" branch
  2. Create a pull request. Assign a reviewer to approve the code changes

Note: Step 2 is required for pushing directly to "master". For a development branch, this step is recommended but not required.

Docstring format

The "Google docstring" format is used in the AMPL code. When writing new code, please use the same Docstring style. Refer here and here for examples.

Versioning

Versions are managed through GitHub tags on this repository.

Built with


Project information

Authors

The Accelerating Therapeutics for Opportunities in Medicine (ATOM) Consortium

Contributors

1. Lawrence Livermore National Laboratory\ 2. GlaxoSmithKline Inc.\ 3. Frederick National Laboratory for Cancer Research\ 4. Computable\ 5. University of California, San Francisco\ 6. Schrodinger\ 7. Leidos  

Support, Suggestions or Report Issues

Contributing

Thank you for contributing to AMPL!

Release

AMPL is distributed under the terms of the MIT license. All new contributions must be made under this license.

See MIT license and NOTICE for more details.