# SGET

**Repository Path**: Tomhappy/SGET

## Basic Information

- **Project Name**: SGET
- **Description**: No description available
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: main
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2026-03-24
- **Last Updated**: 2026-03-24

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

<<<<<<< HEAD
# SGET

## Introduction  
The project  is an implementation of a SGET: Shared-component enhanced Graph convolutional network and Edge-gated Transformer for predicting drug-related microbes.

---

## Catalogs  
- **/data_DM/data**: Contains the dataset used in our method.
- **/code**: Contains the code implementation of  SGET algorithm.
- **dataloader.py**: Processes the drug and microbial similarities, associations, embeddings, and adjacency matrices.
- **sim.py**: Calculates the drug attribute similarity based on the heat kernel and the microbe similarity based on the Gaussian kernel.
- **model.py**: Train and test the model.
- **tools4roc_pr.py**: Contains the early stopping function.

---

## Environment  
The SGET code has been implemented and tested in the following development environment: 

- Python == 3.8,10 
- Matplotlib == 3.7.1
- PyTorch == 2.0.0  
- NumPy == 1.24.1
- Scikit-learn == 1.3.0

---

## Dataset  
- **drug_names.txt**: Contains the names of 1373 drugs.  
- **microbe_names.txt**: Contains the names of 173 microbes.
- **drugsimilarity.zip**: A compressed file that contains the following two files.
  - **drugfusimilarity.txt**: Includes the functional similarities among the drugs.
  - **newdrugsimilarity.txt**: Includes the drug  functional similarities .
- **microbe_microbe_similarity.txt**: Contains the cosine similarity of microbes.
- **newmicrobesimilarity.txt**: Contains the Gaussian kernel similarity of microbes.
- **net1.mat**: Represents the adjacency matrix of the drug-microbe heterogeneous graph.
- **Supplementary_file_SF1.xlsx**: Lists the top 20 candidate microbes for each drug.

---

## How to Run the Code  
1. **Data preprocessing**: Constructs the adjacency matrices, embeddings, and other inputs for training the model.  
    ```bash
    python dataloader.py
    ```

2. **Train and test the model**.  
    
    ```bash
    python model.py
    ```