CellRank for directed single-cell fate mapping

CellRank is a toolkit to uncover cellular dynamics based on Markov state modeling of single-cell data. It contains two main modules: kernels compute cell-cell transition probabilities and estimators generate hypothesis based on these. Our kernels work with a variety of input data including RNA velocity [La Manno et al., 2018] and [Bergen et al., 2020], cellular similarity (both transcriptomic and spatial) and pseudotime, among others. Our VelocityKernel takes into account uncertainty in the velocities and allows you to aggregate the short-range fate relations given by RNA velocity into longer trends along the phenotypic manifold. Our main estimator is Generalized Perron Cluster Cluster Analysis (G-PCCA) [Reuter et al., 2018] which coarse-grains the Markov chain into a set of macrostates which represent initial, terminal and intermediate states. For each transient cell, we compute its fate probability towards any terminal state. We show an example of such a fate map in the figure above, which has been computed using the data of [Bastidas-Ponce et al., 2019]. CellRank combines kernels and estimators with a powerful plotting API, enabling you to visualize e.g. smooth gene expression trends along lineages or fate-informed circular embeddings, to name just a few.

CellRank scales to large cell numbers, is fully compatible with scanpy and scvelo and is easy to use.

Getting started with CellRank

If you’re new to CellRank, make sure to go though the basic tutorial which introduces you to CellRank’s high-level API. Most biological systems require a bit more control, so be sure to check out the kernels and estimators tutorial which allows to unlock the full power of CellRank. If you want to see individual functions in action, visit our gallery.

To use CellRank without RNA velocity information, check out the beyond RNA velocity tutorial as well as the time-series tutorial.

Latest additions

1.4.0 2021-06-30

This release includes:

Additions

• Add new external kernel based on Waddington optimal transport [Schiebinger et al., 2019] cellrank.external.kernels.WOTKernel PR 595.

• Add new tutorial Time series datasets that shows how to use the external Waddington OT kernel PR 35.

• Add reprogramming dataset cellrank.datasets.reprogramming_schiebinger() from [Schiebinger et al., 2019] PR 631.

• Add cellrank.pl.log_odds() to plot log-odds ratio between 2 lineages sorted by experimental time. PR 642.

• Add cellrank.tl.estimators.GPCCA.plot_macrostate_composition() to visualize macrostate composition over a categorical variable PR 641.

• Add cellrank.tl.estimators.BaseEstimator.plot_lineage_drivers_correlation() to plot lineage driver correlation with 2 lineages in a scatter plot PR 640.

• Add cellrank.tl.kernels.Kernel.plot_single_flow() based on [Mittnenzweig et al., 2021] to visualize outgoing transition matrix flow in clusters across experimental time PR 615.

• Dramatically speed-up the computation of fate probabilities (>5x) PR 638.

• Remove 4 technical examples and add 2 new examples PR 602:

  • Plot projection - transition matrix project onto an embedding.

  • Plot random walks - simulation of random walks on a Markov chain.

• Add option to visualize cell-level covariates in cellrank.pl.cluster_lineage() PR 634.

• Add option to force-recompute transition matrix in cellrank.tl.initial_states() and cellrank.tl.terminal_states() PR 577.

• Change cellrank.tl.kernels.PseudotimeKernel defaults and prune available parameters of soft thresholding scheme PR 583.

• Parallelize transition matrix computation in cellrank.tl.kernels.PseudotimeKernel PR 587.

• Prune requirements.txt PR 571.

• Add small improvements to documentation PR 584 PR 601 PR 605 PR 648 PR 35.

Bugfixes

• Fix estimator’s inconsistent state when reading from anndata.AnnData PR 563.

• Fix not checking whether probabilities sum to 1 in cellrank.tl.estimators.BaseEstimator.compute_absorption_probabilities() PR 566.

• Fix always forcing sparse transition matrix in cellrank.tl.kernels.Kernel PR 586.

• Fix passing custom connectivity key in cellrank.tl.kernels.Kernel PR 590.

• Fix kernels in cellrank.external always requiring connectivities PR 600.

• Fix parallelization of sparse matrix with too many jobs PR 633.

• Fix plotting coarse-grained transition matrix when no stationary distribution is found Issue 594.

Manuscript

Please see our preprint on bioRxiv to learn more.

CellRank’s key applications

• compute initial & terminal as well as intermediate macrostates of your biological system

• infer fate probabilities towards the terminal states for each individual cell

• visualize gene expression trends along specific lineages while accounting for the continuous nature of fate determination

• identify potential driver genes for each identified cellular trajectory

Why is it called “CellRank”?

CellRank does not rank cells, we gave the package this name because just like Google’s original PageRank algorithm, it works with Markov chains to aggregate relationships between individual objects (cells vs. websites) to learn about more global properties of the underlying dynamics (initial & terminal states and fate probabilities vs. website relevance).

Support

We welcome your feedback! Feel free to open an issue, send us an email or tweet if you encounter a bug, need our help or just want to make a comment/suggestion.

Contributing

We actively encourage any contribution! To get started, please check out both the contribution guide as well as the external API. CellRank’s modular structure makes it easy to contribute, be it a new method to compute cell-cell transition probabilities (kernels), a new way to analyze a transition matrix (estimators) or an addition to the plotting API. If you’re thinking of contributing a new kernel, we have a kernel tutorial that guides you trough the process.

CellRank was developed in collaboration between the Theislab and the Peerlab.