This repository hosts a mass spectrum prediction framework, designed to facilitate the analysis and prediction of mass spectrometry data. The framework is structured on a divisional approach, consisting of two main components:
- Part Embedding: This component takes an input molecule and generates an embedding representation of it.
- Prediction Head: The head component utilizes the embedding to produce the final prediction.
These components are seamlessly integrated through a combine_model function. The framework is designed to be modular, allowing users to easily add new embeddings or modify the prediction head as per their requirements.
Ensure you have Anaconda installed on your system to manage virtual environments and dependencies.
-
Clone the Repository
git clone [Your Repository URL] cd [Your Repository Name] -
Create a Conda Virtual Environment
Using the
requirements.txtfile provided in the repository, you can create a Conda environment with all necessary dependencies:conda create --name myenv --file requirements.txtReplace
myenvwith your preferred environment name. -
Activate the Virtual Environment
conda activate myenv -
Running the Framework
After activating the environment, you can run the framework:
python main.pyMake sure to modify
main.pyas per your configuration needs.
To customize the framework:
- Adding a New Embedding Part: Insert your code into the embedding module. This allows for new ways to process and embed input molecules.
- Modifying the Prediction Head: Implement your changes in the head module to alter how predictions are made based on the embeddings.
- Configuring the Model: Construct new instances or modify the model configuration in
config_model.py.
Contributions to enhance or extend the framework's capabilities are welcome. Please submit your pull requests or open issues to discuss proposed changes.