DeepMech

DeepMech is an interpretable graph-based deep learning framework for predicting complete chemical reaction mechanisms (CRMs).

---

🚀 Installation

# Clone repository
git clone https://github.com/alhqlearn/DeepMech.git
cd DeepMech

# Create and activate virtual environment
conda create -c conda-forge -n deepmech python=3.6 -y
conda activate deepmech
pip install rdkit-pypi
conda install pytorch cudatoolkit=11.3 -c pytorch -y
conda install -c conda-forge rdkit -y #not required
conda install -c dglteam dgl-cuda11.3 -y
pip install dgllife

# You can also set up the environment using the environment.yml file by running:
`conda env create --name deepmech -f environment.yml`

---

📂 Download Dataset and Model

Download the dataset and model from the following link, and place them in the respective folders:

• Dataset → ./data/ID_Split42
• Model → ./models

🔗 Google Drive Link

If you want to create a new dataset, prepare the .txt files and place them in e.g., ./data/my_data

Required files:

• train.txt
• val.txt
• test.txt

Each file should contain atom-mapped elementary steps in SMILES format, one per line.

---

⚙️ Preprocessing

Run preprocessing scripts inside ./preprocessing:

# Step 1: Extract information from training data
python preprocessing/Extract_from_train_data.py -d ID_Split42

# Step 2: Preprocess dataset
python preprocessing/Run_preprocessing.py -d ID_Split42

---

🏋️ Training

Train your model using:

python scripts/TrainWithAcc.py -d ID_Split42 -m Split42

---

✅ Testing

(a) Single-step Prediction

Run:

python SingleStepPred.py

(b) Full ID-CRM Prediction

1. Train the Reaction Classifier

   Reaction_Classifier/Training_Reaction_Classifier.ipynb

2. Load the Trained Model

   • Update the file reactivity_classifier_AttentiveFP.py with the trained model path.

3. Run Inference

   python AttFP_ID_CRM_infer.py

---

🔮 New Prediction

To predict complete CRMs, specify the input SMILES in the NewCRM_Prediction.py script and execute

python NewCRM_Prediction.py

---

📑 Citation

If you use DeepMech in your research, please cite it as:

@misc{deepmech2025,
  title        = {DeepMech: A Machine Learning Framework for Chemical Reaction Mechanism Prediction},
  author       = {},
  year         = {2025},
  howpublished = {\url{https://github.com/alhqlearn/DeepMech}},
}

---