UniVar is a free, no-login web platform that automates annotation, filtering, and prioritization of genetic variants (SNVs, INDELs, CNVs, SVs) to speed up rare disease diagnosis. No programming skills needed—upload your VCF, add clinical info (or use gene panels if HPO terms are unavailable), and get prioritized candidates in minutes.
The full annotation pipeline (via Ensembl VEP, Nirvana and other tools) may take a few hours to complete depending on the dataset size, but once annotated, variant sorting and filtering finish in seconds, allowing you to quickly review and export results.
Why use it?
- Handles all variant types in one workflow.
- Filters common SVs with our custom, benchmarked catalogue.
- Prioritizes variants effectively even without detailed HPO terms — great for quick screens.
- Benchmarked on real cases, including compound heterozygous SV+SNV detection.
- Secure Access: Uploaded samples are private—only users with the generated shareable link can access results (internal IT team has database access for maintenance).
Read the paper for details.
- Go to the upload page.
- Upload: Drag-and-drop your VCF file (supports SNV/INDEL/CNV/SV).
- Add Context: Select a gene panel or enter HPO terms (optional—UniVar works without!).
- Run Analysis: Hit "UPLOAD" for automated annotation (via Ensembl VEP, Nirvana, etc.) and filtering.
- View Results: Interactive table with scores, pathogenicity predictions, and export to CSV/TSV.
Example Input: Download a sample VCF from upload page and test it.
- Unified Workflow: Annotate and prioritize SNVs/INDELs/CNVs/SVs together.
- Smart Filtering: Population-based SV filtering; allele frequency thresholds.
- Prioritization Options: HPO-based or gene-panel driven (our novel method rivals clinician-assigned terms).
- Outputs: Ranked variant table, visualizations (e.g., IGV), and exports.
- Scalable: Web-first, but local install for heavy use and customization.
- Security: Link-based access control ensures privacy for your data.
Dive deeper in univar-doc:
- User Guide: Step-by-step web usage.
- Advanced Setup: Local deployment.
- Frequently Asked Questions: Frequently Asked Questions.
Start with the univar-doc README for an overview of all available guides.
If you need offline access:
- Clone:
git clone https://github.com/kensung-lab/UniVar.git - Install deps: See docs
Concerned about data privacy? Want to modify the pipelines for your specific needs? Or simply reference our battle-tested Nextflow workflows for alignment and annotation? UniVar's open-source design makes it easy to run everything locally or adapt components to your workflow.
Benefits:
- Privacy-First: Process sensitive genomic data on your own hardware—no uploads required.
- Customization: Tweak parameters, add steps, or integrate with existing tools.
- Reusability: Follow or fork our modular Nextflow pipelines for reproducible results.
Key Components for Customization:
- Alignment-to-Calling Pipeline: End-to-end Nextflow workflow from FASTQ reads to VCFs. Includes alignment (BWA), variant calling (DeepVariant), CNV detection (CNVkit), and SV calling (SurVeyor). Ideal for full WGS/WES processing.
- Annotation Pipeline: Nextflow workflows for comprehensive variant annotation.
- SNV/SV Annotation: Uses Ensembl VEP, VCFanno, and Nirvana for functional and pathogenicity predictions.
- Prioritization with Exomiser: Phenotype-driven ranking.
- Database Setup: Scripts to download and manage annotation databases.
To get started:
- Clone the repo:
git clone https://github.com/kensung-lab/UniVar.git - Follow the installation guide for environment setup (Docker/Nextflow recommended).
- Run pipelines: See subfolder READMEs for
nextflow runcommands and config examples. - Customize: Edit
.nffiles or profiles for your cluster/HPC setup.
For detailed tutorials on pipeline usage and modification, check univar-doc/tutorials.
- univar-doc: Contains the documentation for the Unified Variant Interpretation Platform.
- univar-mongodb: Contains initialization database scripts for the Unified Variant Interpretation Platform.
- univar-data-tools: Contains base data for the Unified Variant Interpretation Platform database.
- hpo-converter: Contains scripts to import HPO Terms into the database.
- import-gene-panel: Contains scripts to import gene panels into the database.
- univar-gene-db: Contains scripts to import gene-related information into the database.
- univar-annotation: Contains the annotation pipeline and related data for the Unified Variant Interpretation Platform.
- annotation-next: A Nextflow workflow for annotating SNVs and SVs using VEP, VCFanno, and Nirvana.
- data-source: Contains scripts to download required annotation databases.
- exomiser-next: A Nextflow workflow for running Exomiser prioritization.
- pipeline-scripts: Contains all pipeline running scripts.
- univar-frontend: Contains the user interface for the Unified Variant Interpretation Platform.
- univar-backend: Contains the backend logic for the user interface of the Unified Variant Interpretation Platform.
- variant-import-tool: Contains the script to convert VCF to MongoDB.
- alignment-2-call-next: Optional. A Nextflow workflow for end-to-end processing from FASTQ files to VCF, including alignment, SNP and indel variant calling with DeepVariant, CNV detection with CNVkit, and SV calling with SurVeyor. Can be replaced with any pipeline that generates VCF files.
- univar-docker: Contains all of the external custom required Dockerfiles for the Unified Variant Interpretation Platform.
UniVar is built on the shoulders of giants in bioinformatics and software development. We gratefully acknowledge the following open-source tools and libraries that power our platform:
- Nextflow: For scalable workflow orchestration.
- DeepVariant: Google's deep learning-based variant caller.
- CNVkit: For copy number variant detection.
- SurVeyor: For structural variant calling from paired-end NGS data.
- Ensembl VEP: Ensembl's tool for SNV and indel variant annotation.
- vcfanno: For flexible VCF annotation.
- Nirvana: For structural variant and CNV annotation.
- Exomiser: For phenotype-driven variant prioritization.
- bwa: For short-read alignment.
- samtools: For manipulating high-throughput sequencing data.
- bcftools: For variant calling and manipulation.
- GLnexus: For joint genotyping from multiple samples.
- MongoDB: For scalable data storage.
- Vue.js & NestJS: For the modern web frontend and backend.
- Additional tools: many more—see the installation guide for full dependencies.
Thank you to all contributors and maintainers for enabling accessible genomic analysis!
- Issues: Open one for bugs/feedback.
- Pull Requests: Welcome! Fork, branch, and submit.
- Community: Join discussions on GitHub Discussions.
Please cite this paper when using Unified Variant Interpretation Platform for your publications.
UniVar: A variant interpretation platform enhancing rare disease diagnosis through robust filtering and unified analysis of SNV, INDEL, CNV, and SV.
Cherie C.Y. Au-Yeung, Yuen-Ting Cheung, Joshua Y.T. Cheng, Ken W.H. Ip, Sau-Dan Lee, Victor Y.T. Yang, Amy Y.T. Lau, Chit K.C. Lee, Peter K.H. Chong, King Wai Lau, Jurgen T.J. van Lunenburg, Damon F.D. Zheng, Brian H.M. Ho, Crystal Tik, Kingsley K.K. Ho, Ramesh Rajaby, Chun-Hang Au, Mullin H.C. Yu, Wing-Kin Sung* (2025).
> https://doi.org/10.1016/j.compbiomed.2024.109560
@article{AUYEUNG2025109560,
title = {UniVar: A variant interpretation platform enhancing rare disease diagnosis through robust filtering and unified analysis of SNV, INDEL, CNV and SV},
journal = {Computers in Biology and Medicine},
volume = {185},
pages = {109560},
year = {2025},
issn = {0010-4825},
doi = {https://doi.org/10.1016/j.compbiomed.2024.109560},
url = {https://www.sciencedirect.com/science/article/pii/S0010482524016457},
author = {Cherie C.Y. Au-Yeung and Yuen-Ting Cheung and Joshua Y.T. Cheng and Ken W.H. Ip and Sau-Dan Lee and Victor Y.T. Yang and Amy Y.T. Lau and Chit K.C. Lee and Peter K.H. Chong and King Wai Lau and Jurgen T.J. {van Lunenburg} and Damon F.D. Zheng and Brian H.M. Ho and Crystal Tik and Kingsley K.K. Ho and Ramesh Rajaby and Chun-Hang Au and Mullin H.C. Yu and Wing-Kin Sung},
keywords = {Variant interpretation, Variant prioritization, Genetic diagnosis, Rare diseases, SNV and INDEL, Copy number variants, Structural variants},
abstract = {Background
Interpreting the pathogenicity of genetic variants associated with rare diseases is a laborious and time-consuming endeavour. To streamline the diagnostic process and lighten the burden of variant interpretation, it is crucial to automate variant annotation and prioritization. Unfortunately, currently available variant interpretation tools lack a unified and comprehensive workflow that can collectively assess the clinical significance of these types of variants together: small nucleotide variants (SNVs), small insertions/deletions (INDELs), copy number variants (CNVs) and structural variants (SVs).
Results
The Unified Variant Interpretation Platform (UniVar) is a free web server tool that offers an automated and comprehensive workflow on annotation, filtering and prioritization for SNV, INDEL, CNV and SV collectively to identify disease-causing variants for rare diseases in one interface, ensuring accessibility for users even without programming expertise. To filter common CNVs/SVs, a diverse SV catalogue has been generated, that enables robust filtering of common SVs based on population allele frequency. Through benchmarking our SV catalogue, we showed that it is more complete and accurate than the state-of-the-art SV catalogues. Furthermore, to cope with those patients without detailed clinical information, we have developed a novel computational method that enables variant prioritization from gene panels. Our analysis shows that our approach could prioritize pathogenic variants as effective as using HPO terms assigned by clinicians, which adds value for cases without specific clinically assigned HPO terms. Lastly, through a practical case study of disease-causing compound heterozygous variants across SNV and SV, we demonstrated the uniqueness and effectiveness in variant interpretation of UniVar, edging over any existing interpretation tools.
Conclusions
UniVar is a unified and versatile platform that empowers researchers and clinicians to identify and interpret disease-causing variants in rare diseases efficiently through a single holistic interface and without a prerequisite for HPO terms. It is freely available without login and installation at https://univar.live/.}
}
Built with ❤️ by the Ken Sung Lab. Questions? Email us.