MOSAIC
Multi-Omics Spatial Atlas in Cancer (MOSAIC), full cohort
Digitized H&E slides, spatial transcriptomics, single-cell and bulk RNA and exome sequencing with clinical data from cancer patients, built to find cancer subtypes, drug targets and biomarkers. 2,725 patients across 11 cancer types, through Owkin's paid K Pro platform.
Overview
MOSAIC (Multi-Omics Spatial Atlas in Cancer) is a multi-center cancer study led by Owkin with academic hospitals in Switzerland, Germany, France and the US. It profiles tumor samples with spatial transcriptomics and single-cell RNA sequencing alongside digitized H&E histology, bulk RNA sequencing, whole exome sequencing and about 100 clinical variables per patient. The goal is a large reference for finding cancer subtypes, drug targets, treatment response markers and biomarkers with machine learning. A 60-patient subset is shared with academic researchers as a separate entry (MOSAIC Window).
Composition
The project website lists 2,725 patients in 11 cancer types: diffuse large B-cell lymphoma, non-small cell lung cancer, ovarian cancer, glioblastoma, bladder cancer, breast cancer, pancreatic cancer, head and neck squamous cell carcinoma, mesothelioma, colorectal cancer and gastric cancer. The per-type counts add up to 2,248, fewer than the total, and the site does not explain the gap. It also gives a count for each data type without saying whether it means patients or samples: 2,666 for H&E whole-slide images, 2,371 for spatial transcriptomics, 2,266 for whole exome sequencing, 2,168 for single-cell RNA sequencing and 2,114 for bulk RNA sequencing. The preprint, written earlier, reports more than 1,900 patients as of April 2025.
Acquisition
The founding partners are Lausanne University Hospital, Charité in Berlin, University Hospital Erlangen, Gustave Roussy near Paris and the University of Pittsburgh. Tumor samples are formalin-fixed and paraffin-embedded. According to the preprint, all participating centers scan the H&E slides on a Leica Aperio CS2 scanner at 40x magnification.
Annotations
No image annotations such as tumor outlines are described. Each patient comes with clinical variables.
Known limitations
The full cohort is not openly available. Owkin offers it in the paid tier of its K Pro platform and publishes no terms for it, so it is unclear what users may download or do with the data. The counts on the website change as the study grows and carry no date.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 2,725 subjects.
Condition
Groups can overlap
- Diffuse large B-cell lymphoma 375 14%
- Non-small cell lung cancer 368 14%
- Ovarian cancer 334 12%
- Glioblastoma 282 10%
- Bladder cancer 265 10%
- Breast cancer 206 8%
- Pancreatic ductal adenocarcinoma 131 5%
- Head and neck squamous cell carcinoma 118 4%
Therapy areas in MOSAIC dataset page (mosaic-research.com/dataset), counts as read on 2026-10-23
License and access
Our reading of the license, not legal advice. Before you use the data, read the original license and confirm that your use is allowed. We take no responsibility for how you use a dataset. Full disclaimer
Access costs money
The full cohort is part of the paid tier of Owkin's K Pro platform, offered through a demo booking or a subscription. Whether it allows downloading the data is not published. A 60-patient subset is available to academic researchers on EGA as MOSAIC Window.
No published license or data use terms
The data holder has published no license and no data use terms. Any use, sharing or commercial right has to be agreed with the data holder, and nothing can be assumed to be allowed. Default copyright and data protection law still apply.
Owkin publishes no terms for MOSAIC data on K Pro. The MOSAIC Window subset has its own data use agreement.
What you can do
- Commercial use Not stated
- Not stated
- Train ML models Not stated
- Create derived data Not stated
- Publish results Not stated
What you can share
- Share the data Not stated
- Share derived data Not stated
- Share trained models Not stated
What you must do
- Cite or credit Not stated
- Share alike Not stated
- Sign an agreement Not stated
- Ethics approval Not stated
- Manuscript review Not stated
- Release code Not stated
- Return results Not stated
- Delete after use Not stated
Limits
- No re-identification Not stated
- Location limits Not stated
Citation
Hoffmann C, MOSAIC Consortium. MOSAIC: Intra-tumoral heterogeneity characterization through large-scale spatial and cell-resolved multi-omics profiling. bioRxiv 2025. doi:10.1101/2025.05.15.654189
Sources
Every number on this page comes from one of these documents. Each chart names the table or page it is taken from. The raw numbers are in stats.csv.