CoolFace
Apppublic

srabhishek/Genomesightpro

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes
App README

GENTIS: Graph-Enhanced Neural Transformer for Integrative Sequence Analysisral Transformer for Integrative Sequence Analysis

![License: MIT](https://opensource.org/licenses/MIT) MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) ![Python 3.8+](https://www.python.org/downloads/)

Overview

GENTIS is a novel computational framework that combines positional sequence analysis, conservation features, and transformer-based machine learning to identify biomarkers across diverse biological datasets. By integrating statistical approaches with deep learning, GENTIS enables more robust biomarker detection in complex genomic data.l framework that combines positional sequence analysis, conservation features, and transformer-based machine learning to identify biomarkers across diverse biological datasets. By integrating statistical approaches with deep learning, GENTIS enables more robust biomarker detection in complex genomic data.

The framework incorporates negative binomial generalized linear models with transformer-based feature extraction to enhance sensitivity while maintaining specificity. Our architecture allows for the automated processing of sequence data, feature engineering, model training, and visualization within a unified computational pipeline, significantly reducing the computational burden of iterative biomarker analysis. binomial generalized linear models with transformer-based feature extraction to enhance sensitivity while maintaining specificity. Our architecture allows for the automated processing of sequence data, feature engineering, model training, and visualization within a unified computational pipeline, significantly reducing the computational burden of iterative biomarker analysis.

Key Features

Framework Architecture

  • —Sequence preprocessing module that handles data cleaning, normalization, and quality control that handles data cleaning, normalization, and quality control
  • —Feature engineering pipeline with both traditional biological feature extraction and deep learning-based approachespeline with both traditional biological feature extraction and deep learning-based approaches
  • —Statistical analysis module incorporating negative binomial generalized linear models for count dataalysis module incorporating negative binomial generalized linear models for count data
  • —Transformer-based sequence analysis using domain-specific pre-trained modelsuence analysis using domain-specific pre-trained models
  • —Integrated biomarker discovery pipeline that combines statistical and AI-based approachesker discovery pipeline that combines statistical and AI-based approaches
  • —Interactive visualization components for result interpretationion components for result interpretation

Data Processing & Analysisysis

  • —Support for DNA/RNA/Protein sequences- Support for DNA/RNA/Protein sequences
  • —Automatic sequence type detectiontype detection
  • —Advanced k-mer analysis
  • —Population and Individual-level analysis-level analysis
  • —Integrated genomics datasetsasets
  • —Real-time sequence validationvalidation

Machine Learning Capabilitieses

  • —Multiple training modes:ing modes:
  • —Basic Training
  • —Hyperparameter Tuning
  • —Ensemble Learning
  • —Custom CNN architecturesom CNN architectures
  • —Transfer learning with transformersrning with transformers
  • —Cross-validation supportrt
  • —Model serialization

Biomarker Discovery### Biomarker Discovery

  • —Automated biomarker detection
  • —Feature importance analysisce analysis
  • —Motif enrichment analysissis
  • —Conservation analysisalysis
  • —Mutation hotspot detection
  • —Functional element predictionlement prediction

Visualizations

  • —Interactive sequence length distributionve sequence length distribution
  • —Nucleotide composition plotside composition plots
  • —K-mer frequency analysisequency analysis
  • —ROC curves
  • —Confusion matriceson matrices
  • —Feature importance plotsure importance plots
  • —Training history visualizationning history visualization
  • —Biomarker network graphsmarker network graphs

Model Evaluation### Model Evaluation

  • —Comprehensive metrics (Accuracy, Precision, Recall, F1)- Comprehensive metrics (Accuracy, Precision, Recall, F1)
  • —ROC curve analysislysis
  • —Cross-validation resultsesults
  • —Model calibration
  • —Prediction distribution analysis- Prediction distribution analysis
  • —Error analysissis

Technologies Used

  • —Python 3.8+hon 3.8+
  • —Streamlit
  • —TensorFlow TensorFlow
  • —Scikit-learn- Scikit-learn
  • —Biopython
  • —Plotly- Plotly
  • —Pandas
  • —NumPy- NumPy
  • —PyTorch & Transformers (for advanced analysis)rs (for advanced analysis)

Installation

Prerequisiteserequisites

  • —Python 3.8 or higher
  • —pip package managerckage manager

Quick Setup Quick Setup

The easiest way to install all dependencies is to use our automated installer script:ll all dependencies is to use our automated installer script:

bash
python install_dependencies.py.py

This script will detect your system configuration and install the appropriate versions of all required packages.your system configuration and install the appropriate versions of all required packages.

Manual Installation

  1. 1.Clone the repository:1. Clone the repository:
bash```bash
git clone https://github.com/your-username/GENTIS.git
cd GENTIScd GENTIS
  1. 1.Create and activate a virtual environment (recommended):ual environment (recommended):
bash
python -m venv venv
bash
# On Windowsn Windows
venv\Scripts\activatevenv\Scripts\activate
bash
# On macOS/Linux# On macOS/Linux
source venv/bin/activatevenv/bin/activate
  1. 1.Install dependencies:3. Install dependencies:
bash```bash
pip install -r requirements.txts.txt

Troubleshooting Dependencies### Troubleshooting Dependencies

If you encounter warnings about missing libraries when running the application:nings about missing libraries when running the application:

bash```bash
# For TensorFlow
pip install tensorflow==2.14.0pip install tensorflow==2.14.0
bash
# For PyTorch and Transformerss
pip install torch==2.0.1 transformers==4.31.0pip install torch==2.0.1 transformers==4.31.0
bashbash
# For UMAP (dimensionality reduction) reduction)
pip install umap-learn==0.5.3

Running the Applicationication

After installation, launch the interactive interface:on, launch the interactive interface:

bash
python start_app.py

Alternatively, you can run the Streamlit app directly:

bash
streamlit run app.py

Example Usage

  1. 1.Load sample data or upload your own FASTA files1. Load sample data or upload your own FASTA files
  2. 2.Explore sequence characteristics through visualization toolsuence characteristics through visualization tools
  3. 3.Configure and train the GENTIS model to identify potential biomarkers3. Configure and train the GENTIS model to identify potential biomarkers
  4. 4.Evaluate model performance and analyze discovered biomarkersformance and analyze discovered biomarkers
  5. 5.Export results and visualizations for further analysis or publicationtion

Project Structure

viral-biomarker-discovery/arker-discovery/
├── data/                # Folder for raw and processed data├── data/                # Folder for raw and processed data
├── models/             # Folder for saving trained models
├── src/               # Source code├── src/               # Source code
│   ├── __init__.pyinit__.py
│   ├── data_processing.py│   ├── data_processing.py
│   ├── model.pyl.py
│   ├── evaluate.py│   ├── evaluate.py
│   ├── visualizations.pyalizations.py
│   └── ai_analysis.py│   └── ai_analysis.py
├── app.py             # Main application filecation file
├── requirements.txt   # Dependencies├── requirements.txt   # Dependencies
└── README.md         # Project documentation    # Project documentation

Contributing

  1. 1.Fork the repository
  2. 2.Create a feature branch (git checkout -b feature/AmazingFeature)2. Create a feature branch (git checkout -b feature/AmazingFeature)
  3. 3.Commit your changes (git commit -m 'Add some AmazingFeature')ges (git commit -m 'Add some AmazingFeature')
  4. 4.Push to the branch (git push origin feature/AmazingFeature)4. Push to the branch (git push origin feature/AmazingFeature)
  5. 5.Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.licensed under the MIT License - see the LICENSE file for details.

Creator

ABHISHEK S R

Copyrightyright

© 2025 GeneAI. All rights reserved.

Acknowledgments

  • —Built with Streamlitreamlit
  • —Uses TensorFlow for deep learningep learning
  • —Employs Scikit-learn for machine learningmploys Scikit-learn for machine learning
  • —Utilizes Biopython for sequence analysisn for sequence analysis

BiomarkerAnalyzer

Overview

BiomarkerAnalyzer is an advanced genomic biomarker discovery platform built with Gradio. It provides a user-friendly interface for analyzing genomic data to identify potential biomarkers associated with various conditions.

Features

  • —Comprehensive Analysis: Perform differential abundance analysis on genomic sequences to identify potential biomarkers.
  • —Interactive Visualizations: Explore your data through volcano plots, heatmaps, fold change distributions, and MA plots.
  • —Pathway Analysis: Understand the biological context of your findings with integrated pathway analysis.
  • —Machine Learning Integration: Leverage machine learning algorithms to evaluate the predictive power of identified biomarkers.
  • —Result Verification: Validate your findings with cross-validation and permutation testing.
  • —Export Options: Export your results in multiple formats (PDF, Excel, CSV, TSV) for further analysis or reporting.

Installation

  1. 1.Clone the repository:
   git clone https://github.com/yourusername/BiomarkerAnalyzer.git
   cd BiomarkerAnalyzer
  1. 1.Install the required dependencies:
   pip install -r requirements.txt

Usage

  1. 1.Run the application:
   python gradio_app.py
  1. 1.Open your web browser and navigate to:
   http://127.0.0.1:7860
  1. 1.Use the interface to:
  2. 2.Load sample datasets or upload your own data
  3. 3.Set analysis parameters
  4. 4.Run the analysis
  5. 5.Explore the results through visualizations
  6. 6.Export your findings

Sample Datasets

The application includes several sample datasets for demonstration and testing:

Synthetic Datasets

  • —Lung Cancer (Synthetic): 12 patient samples and 12 control samples
  • —Breast Cancer (Synthetic): 10 patient samples and 10 control samples
  • —Colorectal Cancer (Synthetic): 8 patient samples and 8 control samples
  • —Prostate Cancer (Synthetic): 15 patient samples and 15 control samples
  • —Melanoma (Synthetic): 9 patient samples and 9 control samples

Public Datasets (Simulated)

  • —TCGA Lung Cancer: Simulated data representing lung cancer samples from The Cancer Genome Atlas
  • —TCGA Breast Cancer: Simulated data representing breast cancer samples from The Cancer Genome Atlas
  • —GEO Colorectal Cancer: Simulated data representing colorectal cancer samples from Gene Expression Omnibus
  • —SRA Prostate Cancer: Simulated data representing prostate cancer samples from Sequence Read Archive
  • —NCBI Melanoma: Simulated data representing melanoma samples from National Center for Biotechnology Information

For more details about the sample datasets, see the sample_data/README.md file.

Data Input Options

  • —Sample Datasets: Choose from pre-loaded sample datasets for quick testing and exploration.
  • —File Upload: Upload your own genomic data files in CSV, TSV, or FASTA format.
  • —Paste Sequences: Directly paste sequence data in FASTA format for analysis.

Analysis Parameters

  • —Statistical Parameters: Set p-value threshold, fold change threshold, and multiple testing correction method.
  • —AI/ML Methods: Choose from various machine learning algorithms and feature selection methods.
  • —Pathway Analysis: Select pathway databases and enrichment methods for biological context.

Output and Visualizations

  • —Results Table: View detailed results of the differential abundance analysis.
  • —Volcano Plot: Visualize the relationship between statistical significance and fold change.
  • —Heatmap: Explore expression patterns of top significant biomarkers.
  • —Fold Change Distribution: Understand the distribution of expression changes.
  • —MA Plot: Examine the relationship between mean abundance and fold change.
  • —Pathway Enrichment: Discover enriched biological pathways associated with identified biomarkers.
  • —ML Feature Importance: Identify which biomarkers are most predictive for classification.
  • —Biomarker Report: Generate a comprehensive report of potential biomarkers with biological context.
  • —Result Verification: Validate findings with cross-validation and permutation testing.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

  • —Developed by ABHISHEK S R
  • —Built with Gradio, Pandas, NumPy, Matplotlib, and Seaborn

GenomeSight Pro: Biomarker Analyzer

A comprehensive platform for biomarker discovery and analysis from genomic data.

Overview

GenomeSight Pro is an advanced genomic biomarker discovery and visualization platform designed for bioinformaticians, researchers, and clinicians. It identifies gene expression, mutation, and sequence biomarkers for cancer diagnosis, prognosis, and treatment selection.

Features

  • —Data Loading: Import data from sample datasets, upload files, or paste sequences directly
  • —Statistical Analysis: Perform differential expression analysis with customizable parameters
  • —Machine Learning Integration: Apply advanced ML techniques for biomarker identification
  • —Pathway Analysis: Explore biological pathways enriched in your dataset
  • —Interactive Visualizations: View your results through multiple interactive plots
  • —Verification: Cross-validate findings for statistical robustness
  • —Report Generation: Create and export biomarker reports in multiple formats

Getting Started

Installation

No installation is required. Simply:

  1. 1.Make sure you have Python 3.8 or higher installed
  2. 2.Double-click the run_biomarker_analyzer.bat file to start the application
  3. 3.A web browser window will open with the application interface

If you encounter any issues with the launcher, you can also start the application by:

python gradio_app.py

Using the Application

  1. 1.Load Data:
  2. 2.Choose a sample dataset, or
  3. 3.Upload patient/control files, or
  4. 4.Paste sequence data directly
  1. 1.Configure Analysis Parameters:
  2. 2.Set statistical thresholds (p-value, fold change)
  3. 3.Choose multiple testing correction method
  4. 4.Configure machine learning and pathway analysis options
  1. 1.Run Analysis:
  2. 2.Click the "Run Analysis" button
  3. 3.Wait for the analysis to complete
  1. 1.Explore Results:
  2. 2.Review interactive visualizations
  3. 3.Examine the biomarker report
  4. 4.Verify statistical robustness
  1. 1.Export Results:
  2. 2.Choose the export format (CSV, Excel, PDF)
  3. 3.Click "Export Results"

Troubleshooting

If you encounter issues running the application:

  1. 1.Python Not Found:
  2. 2.Make sure Python 3.8 or higher is installed
  3. 3.Ensure Python is added to your system PATH
  1. 1.Dependencies Missing:
  2. 2.The application may prompt you to install missing packages
  3. 3.Follow any on-screen instructions to install them
  1. 1.Browser Issues:
  2. 2.If the browser doesn't open automatically, navigate to http://127.0.0.1:7860

License

© 2025 Abhishek S R - All Rights Reserved

Deployment to Hugging Face Spaces

This application can be deployed to Hugging Face Spaces, a platform for hosting machine learning applications. To deploy:

  1. 1.Create a new Space on Hugging Face Spaces.
  2. 2.Clone this repository and push it to the Hugging Face Space repository.
  3. 3.The application will automatically be deployed using the app.py file.

Alternatively, you can use the GitHub workflow in .github/workflows/sync-to-hub.yml to automatically sync your repository with Hugging Face Spaces when you push to the main branch.

Required Files for Deployment

The following files are essential for deployment:

  • —app.py: The main application file optimized for Hugging Face Spaces
  • —pipeline.py: The backend pipeline for analysis
  • —requirements.txt: Dependencies needed for the application
  • —download_data.py: Script to download or generate sample datasets
  • —assets/icon.svg: Application icon for branding

Setting Up on Hugging Face

When setting up your Space, make sure to:

  1. 1.Select "Gradio" as the SDK
  2. 2.Add any necessary secrets
  3. 3.Allocate sufficient resources for the application

Running the Application Locally

To run the application locally, use:

bash
python app.py

For first-time setup, it's recommended to run the data downloader first:

bash
python download_data.py