here comes WebAtlas: our "Google Maps" for tissue atlases with integrated single cell and spatial transcriptomics so anyone & anywhere can access/explore atlas datasets on a web browser. Fantastic collab with Muzlifah Haniffa, led by Tong LI 李彤 Dave Horsfall & Daniela. Links 👇

Paper: Portal where you can access datasets:

The integration of single cell + spatial transcriptomics can build awesome tissue atlases but it is hard to share these datasets in a globally accessible & usable format without technical burden on the users. We aimed to solve this problem with WebAtlas

WebAtlas addresses 2 key challenges in this area: 1) how to deal with myriad technologies and data types in sc/spatial omics? and 2) how to freely browse integrated datasets?

The core of WebAtlas is a data ingestion pipeline that unifies data types from nearly all commonly used atlassing technologies (works on sc/snRNAseq, Visium, Xenium, MERSCOPE, In Situ Sequencing, seqFISH) into the cloud-optimised & scalable Zarr format @zarr_dev

We put special care into image components of spatial data i.e. raw images and cell segmentation masks are handled as OME-Zarr for multi-scale visualisation. This was a parallel effort with SpatialData & we are keen to support their format in the future

We then re-used the awesome Vitessce web visualisation framework from @ngehlenborg that allows fully interactive exploration of sc + ST data. We customised it for coordinated browsing of cell types & gene expression in integrated sc and ST datasets

We showcased WebAtlas on a "triple" modality atlas of the developing human hindlimb from @teichlab that combines scRNAseq, Visium spatial RNA-Seq and In Situ Sequencing (ISS done by @krobertssci in my lab). You can explore it on & few more tweets below.

We harmonised cell types & gene expression across 3X modalities with computational integration: our Cell2location tool to integrate scRNAseq + Visium and StabMap from @shazanfar to transfer cell types & impute gene expression from scRNA-seq to ISS (

The result is stunning WebAtlas (the video on top) where you can cross-query cell types & genes across all modalities in one place, where you can browse the ISS spatial cell map to explore the beautiful structure of the developing bone, muscle and skin across the embryonic limb.

Many fun things to do on the integrated atlas. For example, the scRNAseq -> ISS imputation allowed us to identify a novel spatial gene expression gradient in chondroprogenitors that give rise to the cartilage anlage! On WebAtlas, you can cross-validate this on Visium data!

WebAtlas is tech agnostic and supports many technologies as listed above e.g. see a Xenium dataset of a human breast tumour on WebAtlas here

...and an integrated scRNA-seq + seqFISH atlas of mouse embryogenesis from @MarioniLab ( - webAtlas link here:

Finally, WebAtlas is scalable: to date it has handled scRNAseq with 900k cells and MERSCOPE with 700k cells Currently works best with <10 Million RNA molecules

Try WebAtlas! Our landing page with all datasets tweeted above & shown in paper: Repo: Tutorials:

I am excited about sharing our upcoming @humancellatlas datasets with the community on WebAtlas and a future where we can build "Next-Gen" multi-modal tissue atlases on the cloud. If you are interested, please reach out!

This was a wonderful collab with @Muzz_Haniffa @BioinfoTongLI @Dave_Horsfall @krobertssci @PengHeCam @shazanfar @teichlab. Special thx to @notjustmoore who aligned us with SpatialData from @fabian_theis and @OliverStegle...

..., @zarr_dev & Vitessce @ngehlenborg & @openmicroscopy teams for our technical foundations, and @wellcometrust @sangerinstitute for supporting our work!