Do you want to map cell fate with different data views and pot. large data? Check out CellRank 2 ( https://t.co/ovxiwYMtc6)! The team around myself, @MariusLange8, M. Klein, @fabian_theis were just awarded the "Scientific Originality Prize" as part of the Helmholtz Software Award.

CellRank 2: Unified fate mapping in multiview single-cell data

CellRank 2 is a comprehensive framework for cell fate analysis using large-scale multiview single-cell data. @PhilippWeiler7 @MariusLange8 @dana_peer @fabian_theis https://t.co/HUlHTOfO1j

Very happy to share that CellRank 2 is out in Nature Methods today; this was a fantastic joint effort co-lead between @PhilippWeiler7 and myself, in collaboration with @dana_peer and with strong coding support by Michal Klein ( https://t.co/sdkn11YOxz).

I am really happy and excited to see CellRank 2 out in Nature Methods ( https://t.co/ovxiwYMtc6)! @naturemethods @HelmholtzMunich @CompHealthMuc @TU_Muenchen @MariusLange8 @fabian_theis @dana_peer (1/16)

Don't forget to check out @MariusLange8's tweet motivating our work and outlining the software design!

Finally, how do you know which kernel is appropriate for your data? We summarized the requirements for each as well as how we believe to use each kernel in a decision diagram to guide the kernel choice! See https://t.co/ktXKpV6e2H for detailed tutorials and documentation. (14/14)

Using intestinal organoid data from chase and pulse experiments, we estimate kinetic rates, identify all terminal states compared to competing approaches, and recover regulatory mechanisms. Special thanks to Nicolas Battich for valuable discussions and feedback! (13/14)

The resolution of common time course studies is in the order of hours to days. Alternatively, metabolic labeling of newly transcribed mRNA can yield time-resolved single-cell RNA at a much finer resolution. We devised a method to infer cell fate based on metabolic labels. (12/14)

To highlight the importance of considering within-time-point information, we studied mTEC development. Compared to classical OT, we identified a cluster of putative progenitors and more known drivers of mTEC development. (11/14)

Based on the RealTimeKernel’s transition matrix, we enable studying GEX change in a continuous fashion, and propose an approach for deriving a real-time-informed pseudotime. In comparison, OT-based methods study state change in a discrete manner. (10/14)

To leverage both within and across time point information, we developed the RealTimeKernel: We combine inter-time-point transitions from OT with similarity-based intra-time-point transitions to enable more granular and faithful cell fate mapping. (9/14)

An increasing number of single-cell datasets include experimental time points for which optimal transport (OT) has been used to match cells between time points. However, OT ignores transitions within time points that contain valuable information for directing transitions. (8/14)

In this work, we studied embryoid body development with the CytoTRACEKernel. We automatically recovered terminal states and identified drivers of the endoderm lineage by correlating gene expression with fate probabilities. Thanks to @MaehrLab and Macrina Lobo for input! (7/14)

If a pseudotime cannot be inferred, CytoTRACE @gungulati has been proposed to infer a stemness score but failed on large systems and cannot assign cellular fate. Our CytoTRACEKernel overcomes these limitations by adapting the approach while yielding comparable scores. (6/14)

We introduce the PseudotimeKernel to adapt and generalize the Palantir @SettyM approach to bias edges of a kNN graph. While RNA velocity is not applicable to human hematopoiesis, our new kernel recovers the correct terminal states and corresponding drivers. (5/15)

To take advantage of alternative methods to model cellular change (pseudotime, CytoTRACE), or data views (exp. time points, metabolic labels), we split fate mapping into two steps: Computing cell-cell transition matrices with kernels, and analyzing them with estimators. (4/14)

CellRank 1 combined RNA velocity with gene expression to quantify cellular fate using a probabilistic model formulation in high dimensions. While RNA velocity can give powerful insight, common model assumptions are oftentimes violated ( https://t.co/RsoSK0lpaZ). (3/14)

This work is a joint effort ( https://t.co/9gIGnrptHb) with my co-lead @MariusLange8, and the software wizard Michal Klein https://t.co/8OrURSyUnZ under the supervision of @fabian_theis and @dana_peer. (2/14)

Say hello to CellRank 2, a unified framework for data view independent fate mapping! Have a look at our preprint on bioRxiv ( https://t.co/bJkneo1fZj) and check out @MariusLange8’s tweetorial as well! @HelmholtzMunich @CompHealthMuc @TU_Muenchen @fabian_theis @dana_peer (1/14)