Source-linked AI summary
VizIt: A multi-view framework for exploring single-cell, spatial, and genetic data online
Chenhang Christopher Zhang, Yanqing Lou, Jie Yuan, Mingming Lu, Jacob Parker, Himanshu Chintalapudi, Zechuan Lin, Clemens R. Scherzer, Yuxuan Hu, Ruifeng Hu, Xianjun Dong
TL;DR
Multi-omic exploration is fragmented across modalities even though researchers need complementary biological perspectives. VizIt provides an open-source framework that connects gene-, cell type-, condition-, spatial-, genomic region-, and variant-centered views, demonstrated through customizable Parkinson’s disease atlas datasets. Its scope is interactive exploration of processed outputs rather than computational integration or replacement of upstream analytical software.
Problem
Interactive exploration remains fragmented across modalities and tools, although multi-omic studies require complementary biological perspectives on the same data.
Method
VizIt is an open-source framework that connects multiple biological views across single-cell, spatial, transcriptomic, epigenomic, and genetic data.
Results
VizIt demonstrates customizable multi-omic atlas exploration by organizing heterogeneous datasets through gene-, cell-, spatial-, and genomic-centered modes.
Takeaways & Limitations
The framework can be adapted across research areas without developing a new visualization application for each collection.
Takeaways & Limitations
VizIt examines processed outputs from upstream workflows and does not itself perform computational integration between datasets.
Abstract
from arXiv · showhide
Multi-omic studies increasingly require data to be examined from complementary biological perspectives, yet interactive exploration remains fragmented across modalities and tools. We present VizIt, an open-source framework for multi-view exploration of single-cell and spatial transcriptomic, epigenomic and genetic data. VizIt connects gene-, cell type-, condition-, spatial-, genomic region- and variant-centered views, enabling seamless navigation across biological perspectives. We demonstrate VizIt through the Parkinson's Cell Atlas, a customizable interactive multi-omic resource.
Introduction
Multi-omic exploration is fragmented across modality-specific tools, despite researchers needing complementary biological perspectives. VizIt addresses this gap with an open-source, deployable framework connecting multiple views of processed omics data without requiring users to write code.
- Researchers need to examine multimodal datasets from complementary biological perspectives, but visualization remains fragmented across separate plots, datasets, and modality-specific tools.
- Existing resources improve access to single-cell data but differ in computational requirements, hosting models, and customization options.
- VizIt is an open-source framework for multi-view exploration of single-cell, spatial transcriptomic, epigenomic, and genetic data.
- The framework organizes data around gene-, cell type-, condition-, genomic region-, peak-, and variant-centered views that users can navigate between.
- VizIt can run as a standalone application or customizable institutional web portal, providing interactive access to processed omics datasets without end-user coding.
Multi-view exploration of omics data
VizIt treats transcriptomic, epigenomic, and genetic analyses as interconnected perspectives on shared data. Users can move between biological entities and views to connect cellular patterns with genes, regulatory elements, genomic regions, and variants.
- The framework represents gene, cell type, condition, and cluster questions as complementary views of the same underlying dataset.
- Navigation can connect an observed cellular pattern to its underlying genes or connect an altered gene to relevant cell populations and conditions.
- For single-cell ATAC-seq and multiome data, VizIt supports genomic-signal exploration from peak, gene, cell-type, and genomic-interval perspectives.
- Genetic views support eQTL and caQTL results, reciprocal gene- and variant-centered queries, and display of GWAS statistics alongside molecular associations.
- VizIt links cell type, gene, regulatory element, genomic region, and genetic variant entities, while the available relationships depend on supplied processed data.
Extending multi-view exploration into spatial context
VizIt extends interconnected exploration into spatial transcriptomics by relating molecular patterns to tissue location and anatomical organization. It visualizes processed outputs and complements, rather than replaces, upstream analytical tools.
- The framework supports spatial datasets from platforms including 10x Genomics Visium, Xenium, and MERFISH.
- Spatial views display gene expression and metadata on tissue images or coordinate maps, relating molecular patterns to anatomical compartments and local organization.
- Spatial exploration complements cell- and gene-centered views by examining transcriptional signatures across locations and genes across anatomical and cellular contexts.
- Multiple datasets can be opened side by side to inspect samples, donors, conditions, species, or modalities.
- VizIt examines processed results from upstream workflows rather than performing computational integration or replacing specialized analytical software.
From individual datasets to customizable multi-omic atlases
VizIt is designed as a generalized framework that research groups can deploy for heterogeneous, domain-specific data collections. Its customizable portals support shared navigation across datasets while reducing the technical barrier between analysis and interpretation.
- Processed datasets are served on demand through browser-based rendering of embeddings, genomic tracks, and spatial images.
- A deployment can contain multiple datasets from different assays and studies while providing a common interface for navigating them.
- Organizations can customize the home interface for a research program or biological domain, supporting project-specific and community-facing portals.
- Research groups can deploy VizIt locally or on institutional infrastructure for standalone use, multi-user access, and collaboration.
- The architecture is intended to reduce the technical barrier between computational analysis and downstream biological interpretation.
A common interface for changing biological perspectives
VizIt treats interactive visualization as part of the analytical process, connecting transitions among biological questions and perspectives rather than presenting analysis only at the end.
- A common interface for changing biological perspectives: Interactive visualization supports transitions from disease-associated genes to cell populations, differential expression, regulatory targets, cell types, and spatial distributions.These linked questions motivate movement across complementary biological perspectives.
- A common interface for changing biological perspectives: VizIt was developed to make these transitions straightforward across multi-omic investigations.
- A common interface for changing biological perspectives: Its principal contribution is a multi-view framework connecting biological perspectives, not a new dimensionality-reduction algorithm or statistical method.
processed single-cell, spatial, epigenomic and genetic data through the biological entities researchers
VizIt connects computationally generated results with biological interpretation through a common interface spanning multiple entity-centered perspectives.
- processed single-cell, spatial, epigenomic and genetic data through the biological entities researchers: Investigators can move among gene-, cell type-, condition-, genomic-region-, regulatory-element-, and variant-centered perspectives.
- processed single-cell, spatial, epigenomic and genetic data through the biological entities researchers: The framework provides a common interface between computationally generated results and biological interpretation.
- processed single-cell, spatial, epigenomic and genetic data through the biological entities researchers: The multi-view paradigm is intended for studies combining molecular phenotypes, spatial information, and human genetic associations.
An implementation using ASAP Parkinson’s disease datasets
VizIt was implemented as the Parkinson’s Cell Atlas, organizing heterogeneous Parkinson’s disease datasets into interactive views across samples, genes, cell populations, variants, and genomic loci.
- An implementation using ASAP Parkinson’s disease datasets: The Parkinson’s Cell Atlas is a real-world VizIt implementation built from ASAP Parkinson Cell Atlas datasets.
- An implementation using ASAP Parkinson’s disease datasets: 13 datasets span single-nucleus RNA-seq, single-cell ATAC-seq, 10x Visium, MERFISH, and single-cell eQTL and caQTL summary statistics.
- An implementation using ASAP Parkinson’s disease datasets: The atlas retains gene-, cell-, spatial-, and genomic-centered exploration while organizing heterogeneous datasets in a domain-specific portal.
- An implementation using ASAP Parkinson’s disease datasets: The framework can be adapted to other research areas without developing a new visualization application for each collection.
- An implementation using ASAP Parkinson’s disease datasets: The portal provides UMAP, cell-cluster, gene-expression, association, and genomic-region views for single-cell, spatial, QTL, snRNA-seq, and scATAC-seq data.
- An implementation using ASAP Parkinson’s disease datasets: The same underlying datasets can be examined across samples, genes, cell populations, variants, and genomic loci within a common interface.
- An implementation using ASAP Parkinson’s disease datasets: Representative outputs generated from the deposited datasets are shown in Figure 1.
Availability
VizIt is openly available through archived source code, documentation, and a publicly accessible Parkinson’s Cell Atlas implementation.
- Availability: VizIt is open source, with source code archived at Zenodo and documentation available online.
- Availability: The Parkinson’s Cell Atlas implementation is publicly accessible online.