This repository contains the implementation of a deep learning model for binary classification of chemical compounds represented as SMILES (Simplified Molecular Input Line Entry System) strings. The model architecture combines convolutional neural networks (CNNs) and long short-term memory (LSTM) networks to identify bioactive compounds for antimicrobial activity (Include the microbial specie the model was trained for). Additionally, it provides tools for interpreting the model's predictions, such as generating Class Activation Maps (CAMs) to highlight important structural motifs.

• Data Preprocessing (Data_processing1.py and Data_processing2.py): Scripts for cleaning SMILES data, including filtering, standardizing, and dataset partitioning.
• Data Loader (data_loader.py): Implements a PyTorch Lightning DataModule for efficient data handling and batching.
• Model (model.py): Defines the neural network architecture combining convolutional, LSTM layers, and multi-head attention, focusing on learning from the structural patterns in SMILES data.
• Training (trainer.py): Facilitates model training with checkpoints, hyperparameter configuration, and TensorBoard logging.
• Attention_Viz.ipynb`) and Grad_CAM_Viz.ipynb: Jupyter Notebook for visualizing activation maps on molecular structures, offering insights into the model's focal points.

To set up the project, follow these steps:

1. Clone the Repository

   git clone https://github.com/catenate15/CAMbiotics
   cd your-repository

2. Install Dependencies Ensure you have Python 3.8+ installed, then run:

   pip install -r requirements.txt

To use the project, follow these steps:

1. Initial Preprocessing:

   • Run Data_processing1.py to clean and standardize SMILES data.

     python Data_processing1.py <path_to_raw_smiles.csv>

2. Further Data Preparation:

   • Execute Data_processing2.py for additional preprocessing, including filtering based on structural features and dataset splitting.

     python Data_processing2.py <path_to_standardized_smiles.csv>

• Train the model using trainer.py, specifying the path to the preprocessed dataset and desired hyperparameters.

  python trainer.py --data_path <path_to_preprocessed_data.csv> --epochs 100 --batch_size 64

• To visualize the model's focus through activation maps on molecular structures, run the Grad_CAM_Viz.ipynb and Attention__Viz.ipynb notebook in Jupyter: You can replace with your own SMILES. Ensure to load the pretrained model

  jupyter notebook Grad_CAM_Viz.ipynb

  jupyter notebook Attention_Viz.ipynb

• Follow the instructions within the notebook to load a trained model checkpoint and visualize activation maps on specified SMILES strings.

Contributions are welcome. Please adhere to the project's coding conventions and commit guidelines. Open a pull request with detailed descriptions of your changes or enhancements.