Quick Start to use the CLI

This section provides instructions for using the command-line interface based on the official code repository: https://github.com/SeulLee05/READRetro

Installation

Follow these steps to set up your environment and install READRetro.

1. Set Up Conda Environment (Recommended)

Create a new Conda environment specifically for READRetro to avoid conflicts with other Python packages:

conda create -n readretro python=3.8 -y
conda activate readretro

2. Install Dependencies and READRetro Package

  • Option A: Install via Pip (includes core dependencies):

    This is the simplest way to get started.

    pip install readretro==1.2.0

    Note: This pip installation attempts to install core dependencies. You might still need to manually ensure PyTorch with CUDA support is correctly installed if you plan to use a GPU. To install PyTorch with specific CUDA version (e.g., 11.3):

    conda install pytorch==1.12.0 cudatoolkit=11.3 -c pytorch
# Then install other dependencies if not covered by readretro pip package
pip install easydict pandas tqdm numpy==1.22 OpenNMT-py==2.3.0 networkx==2.5
conda install -c conda-forge rdkit=2019.09 # Ensure RDKit is compatible
  • Option B: Install from Source (for development or specific modifications):

    Clone the repository and install dependencies.

    git clone https://github.com/SeulLee05/READRetro.git
cd READRetro
# Install dependencies as listed in Option A (PyTorch, RDKit, others)
# or use a provided requirements.txt if available in the repository
# pip install -r requirements.txt

Data Setup (READRetro_data)

The READRetro models and evaluation scripts require specific data files, including pre-trained model weights, datasets for evaluation, and scripts.

  1. Download Data: Download the READRetro_data folder from Zenodo: https://zenodo.org/records/11485641.

  2. Directory Structure: The expected directory structure within READRetro_data is:

    READRetro_data/
├── data.sh
├── data/
│   ├── model_train_data/
│   └── multistep_data/
├── model/
│   ├── bionavi/
│   ├── g2s/
│   │   └── saved_models/
│   ├── megan/
│   └── retroformer/
│       └── saved_models/
├── result/
└── scripts/

  3. Place and Prepare Data

    • Move the downloaded READRetro_data folder into your main READRetro project directory (e.g., if you cloned the GitHub repo, it should be READRetro/READRetro_data).

    • Navigate into the READRetro_data directory and run the setup script:

    cd READRetro_data
sh data.sh
cd ..  # Go back to the main READRetro directory

  4. Verify Data Paths

    Ensure that the symbolic links or copied data are correctly pointing to the locations expected by the scripts.

    • READRetro/retroformer/saved_models should correspond to READRetro_data/model/retroformer/saved_models.

    • READRetro/g2s/saved_models should correspond to READRetro_data/model/g2s/saved_models.

    • READRetro/data should correspond to READRetro_data/data/multistep_data.

    • Other paths like result/ and scripts/ should also align.

    The directories READRetro_data/model/bionavi/, READRetro_data/model/megan/, and READRetro_data/data/model_train_data/ are typically needed for reproducing results from the original manuscript.

Model Preparation

  • Using Pre-trained Models

    The READRetro_data bundle from Zenodo includes pre-trained model checkpoints for Retroformer and Graph2SMILES, which are placed in the correct directories by the data.sh script. These are generally located under READRetro/retroformer/saved_models/ and READRetro/g2s/saved_models/.

  • Training Your Own Models

    If you wish to train your own models, refer to the official repositories for

    The demo.ipynb often included in the READRetro repository provides more detailed instructions on training and using custom models. Ensure your custom model checkpoints and vocabulary files are placed where the READRetro scripts expect them.