Vignet is an interactive web platform for exploring vaccine-gene associations extracted from biomedical literature using natural language processing and machine learning. It enables researchers to discover vaccine mechanisms, identify therapeutic targets, and analyze cross-vaccine gene networks at scale.
Vignet answers critical questions in vaccine research:
- What genes does a vaccine interact with and through which biological pathways?
- How do different vaccines compare in their gene targets and mechanisms?
- Which vaccines are associated with a gene set from my research?
- What does the literature say about specific vaccine-gene interactions?
Built on 240,127 PubMed articles and the Vaccine Ontology, Vignet combines literature mining, interactive visualization, and AI-powered analysis to accelerate vaccine research and drug discovery.
Live Site: https://ignet.org/vignet/
- Explore Vaccines — Search and browse 638 vaccines with mention counts and gene associations
- VacNet — Interactive network visualization of vaccine-gene-drug-disease relationships
- VacPair — Query specific vaccine-gene pairs for co-occurrence evidence and prediction scores
- Gene Enrichment — Upload a gene list and discover associated vaccines in literature
- Compare Vaccines — Side-by-side analysis with shared vs. unique gene targets (Venn diagrams)
- VacSummarAI — AI-powered summaries of vaccine-gene literature with follow-up Q&A
- VO Ontology Browser — Navigate the complete Vaccine Ontology hierarchy with gene data indicators
- Vaccine Assistant — Natural language Q&A about vaccine-gene interactions grounded in evidence
- Analyze Text — Detect genes/vaccines in biomedical text and predict interactions via BioBERT
- Generate Report — Create downloadable HTML reports analyzing vaccine associations for gene lists
Frontend:
- React 19, Vite 8, JavaScript (JSX)
- Tailwind CSS 3 for styling
- Cytoscape.js for network visualization
- Recharts for data visualization
Backend:
- Flask + Waitress (Python 3.12)
- RESTful API at
/api/v1/ - MariaDB database (10.5+)
Machine Learning & NLP:
- BioBERT for protein interaction prediction
- SciBERT for biomedical named entity recognition
- GPT-4o for AI summarization and Q&A
Data:
- PubMed abstracts (240K+ articles, 2025 coverage)
- Vaccine Ontology (VO) — 6.8K nodes, 638 vaccine terms
- DrugBank — drug target interactions
- HDO (Human Disease Ontology) — disease terms
Vignet uses the shared ignet MariaDB database with these core tables:
| Table | Records | Description |
|---|---|---|
t_vo |
586K | Vaccine mentions from PubMed with sentence context |
t_vo_hierarchy |
6.8K | Vaccine Ontology tree structure |
t_vo_has_gene_data |
638 | Lookup table: which VOs have gene associations |
t_gene_pairs |
5.1M | Gene co-occurrence pairs with BioBERT scores |
t_gene_sentence_pairs |
2.4M | Sentence-level gene pair evidence |
Note: The database schema is shared with Ignet and maintained in the Ignet repository.
Access Vignet data programmatically at https://ignet.org/api/v1/
Key Endpoints:
GET /vaccine/search?q=covid— Search vaccinesGET /vaccine/{vo_id}/genes— Get genes for a vaccinePOST /pair/search— Find vaccine-gene co-occurrence evidencePOST /enrichment/analyze— Run enrichment on gene listPOST /text/analyze— Detect genes/vaccines in text
Documentation: https://ignet.org/api/docs/
Access: Public and open; no authentication required. Rate limited to 100 requests/minute.
Integrate Vignet with Claude and other AI systems via Model Context Protocol:
https://ignet.org/api/v1/mcp
Enables natural language queries about vaccine-gene interactions with automatic evidence citation and summarization.
- Node.js 18+ and npm 9+
- Python 3.12+
- MariaDB 10.5+ (shared with Ignet)
cd frontend
npm install
npm run dev # Dev server on http://localhost:5174
npm run build # Production build to ../dist-react/# Assuming the shared Ignet Flask API is running locally
# See Ignet repository for backend setup (port is configurable)
# Set VITE_API_PROXY_TARGET in frontend/.env to point at your backend,
# for example:
# VITE_API_PROXY_TARGET=http://127.0.0.1:5000
# Test connectivity via the public API
curl https://ignet.org/api/v1/vaccine/search?q=covidThe database schema is defined in the Ignet repository:
# Clone Ignet to get schema
git clone https://github.com/hurlab/Ignet.git
# Import schema: scripts/schema_ignet.sqlVignet is deployed as a static SPA served by Apache at https://ignet.org/vignet/,
with the Flask/Waitress backend (shared with Ignet) serving /api/v1/ on the same host.
# Build static assets
cd frontend
npm run build # outputs to ../dist-react/Deploy the contents of dist-react/ to your web server. The frontend calls a
same-origin /api/v1/ path, so no runtime configuration is required in production.
# Target for the dev proxy; defaults to http://127.0.0.1:5000 if unset
VITE_API_PROXY_TARGET=http://127.0.0.1:5000
- User Manual — Complete guide to using all Vignet tools
- API Documentation — REST API reference
- MCP Integration — AI assistant integration guide
- Database Schema — Complete schema definition
- Ignet — General gene-gene interaction networks (sister project, shares database)
- VIOLIN — Vaccine Information and Ontology Linked kNowledgebase
- Vaccine Ontology — Official VO documentation
Vignet integrates:
- PubMed: 240,127 articles, 586K vaccine-gene mentions mined through 2025
- Vaccine Ontology (VO): 638 vaccine terms organized in 6.8K node hierarchy
- BioBERT: Machine learning for protein interaction prediction
- DrugBank: Drug-gene-disease interactions
Cite Vignet as:
Vignet: A vaccine-focused integrative gene network database from
PubMed literature mining. Junguk Hur (University of North Dakota)
and Yongqun "Oliver" He (University of Michigan), 2025-2026.
Supported by NIH/NIAID U24AI171008 VIOLIN 2.0: Vaccine Information
and Ontology Linked kNowledgebase.
BibTeX:
@database{vignet2025,
author = {Hur, Junguk and He, Yongqun},
title = {Vignet: A vaccine-focused integrative gene network database},
year = {2025},
url = {https://ignet.org/vignet/},
note = {Supported by NIH/NIAID U24AI171008}
}This project is supported by:
- NIH/NIAID: U24AI171008 — VIOLIN 2.0: Vaccine Information and Ontology Linked kNowledgebase
- University of North Dakota: Computational Biology & Bioinformatics
- University of Michigan: Medical School & School of Information
This project is released under the MIT License.
Vignet data is derived from publicly available PubMed abstracts and ontology resources. Data is provided as-is for research and educational purposes.
Authors:
- Junguk Hur, Ph.D. — University of North Dakota
- Yongqun "Oliver" He, Ph.D. — University of Michigan
Database Development:
- PubMed text mining with SciMiner
- BioBERT model training and deployment
- Vaccine Ontology curation and integration
Acknowledgments: See Full Acknowledgments
- Website: https://ignet.org/vignet/
- FAQs: https://ignet.org/vignet/faqs/
- Contact: hurlabshared@gmail.com · https://ignet.org/vignet/contact/
- Report Issues: https://github.com/hurlab/Vignet/issues
Last Updated: April 14, 2026 Version: 1.0.1 Copyright © 2025–2026 Vignet. Developed by Hur Lab (UND) & He Lab (UM).