Integrative Multi-Omics and Blood Routine Analysis for Risk Prediction

Figure 1: Overall Project Framework

---

Table of Contents

• Project Overview
• Dependencies
• Code Structure
• Usage Instructions
• File Descriptions
• Execution Examples
• Results
• Contact Information

---

Project Overview

This project focuses on early prediction of chronic disease risks by integrating blood routine data and multi-omics data using deep learning and clustering methods.

Objectives:

• Perform clustering analysis on blood routine data to discover hidden patterns.
• Use deep learning to analyze multi-omics data and identify critical features.
• Build a classification model for disease risk prediction based on blood routine data.

---

Dependencies

Ensure the following dependencies are installed:

• Python 3.9+
• PyTorch 2.5+
• Numpy 2.0+
• Scikit-learn
• Matplotlib

---

Code Structure

project-root/
│-- blood clustering-real.py   # Risk classification using blood routine data
│-- datasets.py                # Data loading and preprocessing
│-- lable_gen.py               # Clustering analysis for blood routine data
│-- load_data.py               # Data loading utilities
│-- main.py                    # Multi-omics feature extraction and training
│-- network.py                 # Deep neural network architecture
│-- Nmetrics.py                # Evaluation metrics
│-- utils.py                   # Utility functions

---

Usage Instructions

1. Clustering Analysis for Blood Routine Data

To analyze and cluster blood routine data, run:

python lable_gen.py

---

2. Multi-Omics Feature Extraction and Training

To train the model and identify critical features from multi-omics data:

python main.py

---

3. Clinical Risk Classification

To perform disease risk classification on blood routine data:

python blood clustering-real.py

---

File Descriptions

File	Description
blood clustering-real.py	Performs disease risk classification on blood routine data.
lable_gen.py	Conducts clustering analysis of blood routine samples.
main.py	Trains the model and extracts key features from multi-omics data.
network.py	Defines the neural network architecture.
datasets.py	Loads and preprocesses blood routine and multi-omics datasets.
Nmetrics.py	Provides evaluation metrics like accuracy, precision, recall, and F1-score.
utils.py	Includes helper utilities for visualization, logging, etc.

---

Execution Examples

Run Clustering Analysis

To cluster blood routine data:

python lable_gen.py

---

Train Multi-Omics Model

To analyze multi-omics data and identify key features:

python main.py

---

Risk Classification

To predict disease risks using blood routine data:

python blood clustering-real.py

---

Results

• Clustering Analysis: Outputs clustering labels and visualizations for blood routine data.
• Feature Identification: Saves extracted key features from multi-omics data.
• Risk Classification: Outputs predicted risk levels along with evaluation metrics (accuracy, precision, recall, and F1-score).

Results are saved in structured formats for further analysis.

---

Contact Information

For inquiries, please contact: Email: dzb20@nudt.edu.cn