CellRank: Advanced Cellular Fate Mapping Now Available on Vecura
This update enables computational biologists and researchers to map cell differentiation trajectories through a guided, streamlined workflow inside Vecura, bypassing the need for complex local infrastructure and manual environment configuration.
What is CellRank?
CellRank 2 is a powerful, unified Markov-state-modeling framework designed to study cellular dynamics in single-cell RNA-seq and multi-omic data. Rather than relying on a single signal, it integrates diverse biological priors—such as RNA velocity, pseudotime, CytoTRACE, or experimental time points—to predict the differentiation trajectories of individual cells. By utilizing Generalized Perron Cluster Cluster Analysis (GPCCA), it enables researchers to identify coarse-grained macrostates and compute fate probabilities, offering a robust, uncertainty-aware approach to understanding complex biological systems.
It helps users map out where cells are headed during development and identify the key genetic drivers behind these fate decisions. It is especially useful for developmental biologists and computational researchers analyzing lineage commitment and cell differentiation processes in high-throughput single-cell datasets.
What can users do with CellRank on Vecura?
With CellRank on Vecura, users can:
- Construct directed transition models: Build probabilistic maps of cell-cell movement using pluggable kernels tailored to their specific experimental data.
- Infer complex lineage structures: Automatically identify initial, intermediate, and terminal states, and calculate the likelihood of each cell reaching a specific fate.
- Interpret lineage biology: Rank genes by their correlation with specific differentiation paths to identify putative drivers of cell-fate decisions.
- Fit expression trends: Model gene expression changes over pseudotime to visualize and validate how gene programs evolve along different lineages.
What the output means
The output provides comprehensive annotations, including a directed transition matrix, identified macrostates and terminal-state classifications, per-cell fate probability matrices, and statistical rankings of lineage driver genes. Furthermore, users receive fitted gene-trend models that illustrate expression dynamics with confidence bands.
This output should be used to support scientific decision making by providing actionable hypotheses regarding cellular differentiation. It does not replace experimental validation and should be interpreted as a statistical model of the underlying biological process.
Why this matters
Understanding how stem or progenitor cells differentiate into mature cell types is fundamental to developmental biology, regenerative medicine, and disease modeling. However, inferring these processes from static single-cell snapshots is inherently challenging due to noise, technical limitations, and the dynamic nature of gene regulation.
CellRank addresses these challenges by providing a principled, unified framework that combines multiple biological signals. By reducing sensitivity to noisy data through kernel blending and providing rigorous uncertainty estimates via GPCCA, it enables researchers to extract more reliable insights from complex scRNA-seq datasets, ultimately accelerating the discovery of regulatory networks governing cell fate.
- Developed by: The scverse community and researchers at Helmholtz Munich.
- Source: Official GitHub repo
- Reference: CellRank 2 paper (Nat Methods 2024)
Vecura で CellRank 2 を試す。
モデルワークスペースを開き、ご自身の入力で評価を始めましょう。