PromptSL

PromptSL is a prompt-learning framework based on Biomedical Large Language Models (LLMs) for context-specific synthetic lethality (SL) prediction.

Overview

File Structure

PromptSL
├── data
│   ├── dataset_example
│   ├── KG
│   │   └── kge.py          
│   ├── omics_data          
│   ├── text_template       
│   └── utils.py            
├── src
│   ├── __pycache__         
│   ├── data_loader.py      
│   ├── data_prepare.py     
│   ├── evaluate.py         
│   ├── main.py             
│   ├── model.py            
│   ├── preprocess.py       
│   └── train.py            
├── environment.yml         
├── LICENSE                 
├── PromptSL_overview.png   
└── README.md               

Requirements

To set up the environment for PromptSL, use the provided environment.yml file. You can create the necessary Conda environment with the following commands:

conda env create -f environment.yml
conda activate promptsl

A CUDA-enabled GPU is required to run the code efficiently.

Training and Evaluation

Training

To train the PromptSL framework, run the main.py script. You can specify various hyperparameters using the command line or configuration files. Ensure that the paths to the datasets are correctly set in the main.py file. There is an example of datasets available at link.

Example command:

python -m src.main --cell_line all --predictor_lr 5e-4 --encoder_lr 1e-4 --projector_lr 1e-4 --om_encoder_lr 5e-4 --batch_size 16 --num_epochs 50 --early_stop_patience 6 --lr_patience 3 --weight_decay 1e-4 --device 0 

Evaluation

To evaluate the trained model, use the evaluate.py script.

Example command:

python -m src.evaluate 

The training results, logs, and saved models, as well as the evaluation results, will be saved in the ./save folder. Ensure that the paths to the datasets and models are correctly set in the evaluate.py file.