We present #Zebrahub: a timecourse atlas of zebrafish embryonic development, combining #scRNAseq time-course data with #lightsheet live imaging. Explore our seq. and imaging datasets interactively at 1/n

All datasets, instrument designs, software packages, protocols are accessible for download at to facilitate the reproducibility and accessibility of our research. We are uploading some datasets & cleaning repos, please be patient with us ;-) 2/n

The scRNAseq dataset consists of a single-embryo high-quality #scRNAseq timecourse dataset for zebrafish post-gastrulation development (10, 12, 14, 16, 19, 24 hpf). In addition, we also provide 2, 3, 5, and 10 days post-fertilization data. 3/n

Our optimized single-embryo cell dissociation protocol lets us do a detailed analysis of gene expression variance across sibling embryos. Interestingly, the gene expression distributions across siblings reach a point of minimal gene expression variability at 19 hpf. 4/n

The scRNAseq data captures the main temporal and anatomical developmental features of vertebrate embryogenesis. From the establishment of the nervous system to mesenchyme lineages. We identified a total of 200 cell types across all 10 developmental stages. 5/n

We also applied the RNA velocity method to our data and obtain a striking 3D projection (UMAP) of these vector flows in the context of our cell type annotations. The data is complex and fascinating, we are only looking at the tip of the iceberg! 6/n

You can explore the #scRNAseq datasets and cell type annotations interactively here: 7/n

#Zebrahub is not only about #scRNAseq data. We also imaged zebrafish embryonic development using two complementary #lightsheet microscopes (#OpenSiMView & #DaXi) to be able to follow as accurately as possible tens and hundreds of thousands of cells in embryos. 8/n

in the team developed a novel cell tracking algorithm called #ultrack to perform embryo-scale cell #segmentation and #tracking. You can find it here: together with all other software developed or used for this work. 9/n

In this video, we track a single cell until it divides into two daughter cells that we keep following! We provide the cell tracking data for the timelapse image data. With this data, you have a digital embryo that can be used to do 'virtual experiments'. 10/n

We performed 'in silico' fate mapping experiments: we marked cells of interest and followed them over time... With a #digital_embryo, you can do this as many times as you need - which is key to understanding the embryo's incredibly complex multicellular flows. 11/n

The best part is that you can do such experiments at home too using our @napari_imaging plugin and downloading our data. Check it out here: - we are still ironing out some kinks here and there. Let us know if you face issues! 12/n

Using #Zebrahub, we studied for the first time the state transition of a key population of late pluripotent axial progenitors called neuromesodermal progenitors (#NMP), challenging classical developmental biology. 13/n

We noticed in the #scRNAseq data how strikingly central the #NMPs are. Zooming into our RNA velocity we can see indeed two branches going to the neural and mesodermal lineages. Our data even suggest a transcriptomic connection to the notochord and floor plate! 14/n

Going deeper we find that our RNA velocity data and a pseudotime analysis suggest that NMPs are pluripotent (give rise to mesodermal and neural progeny) only during early axis elongation before having their fate restricted to the mesoderm. 15/n

Using the high-resolution image-based cell tracking together with our @napari_imaging plugin for in silico fate mapping we confirmed the NMP state transition in pluripotency. 16/n

'in silico' fate mapping is great, but there is nothing better than a real experiment. Next, we performed similar in vivo photo-manipulation experiments directly in our multi-view light-sheet microscope. Again we find the same fate restriction. 17/n

Using the in vivo experiment we can reconstruct the track of a single pluripotent presumptive #NMP! 18/n

To understand the relevant tissue kinematics, we applied the recently developed dynamic morphoskeletons framework from @Mattia__Serra et al. and found that pluripotent NMPs coincide with a strong repeller structure. For more details on the theory check the preprint 😉 19/n

We are still trying to understand all this, and have some ideas in the preprint. We expect and hope for a robust and enthusiastic discussion with the community and welcome feedback, ideas, and suggestions! Reach out at [email protected] and [email protected] 20/n

Thanks to last-minute help from @stardazed0 and the admirably twitter-less Jeremy Maitin-Shepard we managed to integrate in Check it out! It's amazingly fast! 21/n

This work would not have been possible without the amazing #team at @czbiohub sf, in particular, @Merlin_Lange who led this work, and a special shoutout to @ale_agranados, @Shruthi94Vijay, @jobragantini, and @Sarah_E_Ancheta for there wonderful contribution. 22/n

All the @czbiohub contributors @mikeborjatweets, Sheryl Paul, Honey Mekonen, Angela Detweiler, @liilii_tweet, Erin McGeever, @Bin_YANG_Optics, @hoover_zhao, @yang_gp, @kyleawayan Samuel D’Souza, @Adrian_Jacobo, @keirballa, Rafael Gómez-Sjöberg, Greg Huber, Norma Neff, @drAOPisco

"Plus to our wonderful collaborators, Olivier Pourquie, @Mattia__Serra, @SreejithS_, @haesleinhuepf, @AlexandreDizeux 24/n

Many thanks to the @slschmid_CZB, Joe DeRisi, and @StephenQuake for mentorship in the past 5 years! and to the donors of the @czbiohub for their support.

🐣So, on a much lighter note, it is the week end after all! I have hidden in the website 4 easter eggs… 👾🎮 Whoever finds it first wins a zebrahub teeshirt (exclusive for the team but you get one) sent whenever you are!