Merck / MolPROP

fuses molecular language and graph representation for property prediction

MolPROP: Molecular Property Predictions with Multimodal Language and Graph Fusion

fuses molecular language and graph representation for property prediction

current models include:

• language -- ChemBERTa-2-77M-MLM, ChemBERTa-2-77M-MTR
• graph -- GCN, GAT, GATv2

@article{rollins2024,
        title = {{MolPROP}: {Molecular} {Property} {Prediction} with {Multimodal} {Language} and {Graph} {Fusion}},
        journal = {Journal of Cheminformatics},
        author = {Rollins, Zachary A and Cheng, Alan C and Metwally, Essam},
        month = may,
        year = {2024}}

requirements

• git lfs (for locally stored language models)
  • install git lfs for your machine: https://github.com/git-lfs/git-lfs
  • chemberta requires local installation to config: https://huggingface.co/DeepChem
• molprop.yaml

    conda env create --name molprop --file molprop.yaml

preprocess data

data splitting

• data splits to reproduce manuscript results are provided: ./data/*csv
• data should be in data directory (e.g., csv format) and organized such that SMILES string and endpoints are in separate columns
• data can be split using the Bemis-Murko scaffold split implemented by DeepChem
• explore the help options to execute the data split procedure

    python ./data/split.py --help

train and hyperparameter tune

• training and tuning execution is specified by the configuration files (config/setup.json)

    python ./src/train_tune.py

hyperparameter tune

• setup["training"]["ray_tune"] == True
• specify hyperparameter search space in the 'main' of ./src/train_tune.py
• ray cluster must be initialized before hyperparameter tuning execution
• submit PBS script with specified num_cpus and num_gpus
• start ray cluster

    ray start --head --num-cpus=8 --num-gpus=4 --temp-dir="/absolute/path/to/temporary/storage/"
    python ./src/train_tune.py

inference and holdout

• test trained/validated models on holdout by specifying config/setup.json
• MolPROP models trained on data in manuscript are located in models/weights
• setup['holdout']['model_path']
• setup['holdout']['holdout_data_path]

    python ./src/holdout.py

logging information and model storage

• train_tune.log files are recorded and saved for every time stamped batch run
• runs are also recorded on tensorboard
• ***** = unqiue file identifier (e.g., time stamp or number)

    logs/batch_*****/train_tune.log
    logs/batch_*****/events.out.tfevents.***** (setup["training"]["ray_tune"] == False)
    logs/batch_*****/ray_tune/hp_tune_*****/checkpoint_*****/events.out.tfevents.***** (setup["training"]["ray_tune"] == True)

• hyperparameter tune runs are implemented by ray tune and models are stored
• non-hyperparameter tuned models are also stored

    logs/batch_*****/ray_tune/hp_tune_*****/checkpoint_*****/dict_checkpoint.pkl (setup["training"]["ray_tune"] == True)
    models/weights/batch_*****/MOLPROP*****.pth (setup["training"]["ray_tune"] == False)

License

MolPROP fuses molecular language and graph for property prediction.
Copyright © 2023 Merck & Co., Inc., Rahway, NJ, USA and its affiliates. All rights reserved.

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program.  If not, see <http://www.gnu.org/licenses/>.