Welcome to cell2location’s documentation!

For installation instructions see: https://github.com/BayraktarLab/cell2location#Installation

For FAQ and to ask any questions please use scverse discourse: on cell2location https://discourse.scverse.org/c/ecosytem/cell2location/42, on scvi-tools https://discourse.scverse.org/c/help/scvi-tools/7 or in Visium data https://discourse.scverse.org/c/general/visium/32.

For reporting bugs or other issues with cell2location please use GitHub Issues: https://github.com/BayraktarLab/cell2location/issues

Quick start tutorial:

• Mapping human lymph node cell types to 10X Visium with Cell2location
  • Workflow diagram
  • Contents
  • Loading packages
  • Loading Visium and scRNA-seq reference data
  • Estimation of reference cell type signatures (NB regression)
  • Cell2location: spatial mapping
  • Visualising cell abundance in spatial coordinates
  • Downstream analysis
    • Identifying discrete tissue regions by Leiden clustering
    • Identifying cellular compartments / tissue zones using matrix factorisation (NMF)
    • Estimate cell-type specific expression of every gene in the spatial data (needed for NCEM)
  • Advanced use
    • Working with the posterior distribution and computing arbitrary quantiles
    • Modules and their versions used for this analysis

Cell2location package is implemented in a general way (using https://pyro.ai/ and https://scvi-tools.org/) to support multiple related models - both for spatial mapping and estimating reference cell type signatures:

1. Cell2location for spatial mapping of cell types which estimates cell abundance by decomposing spatial data into reference expression signatures of cell types (LocationModelLinearDependentWMultiExperimentLocationBackgroundNormLevelGeneAlphaPyroModel).

2. Models for estimating reference expression signatures of cell types from scRNA data, accounting for variable sequencing depth between batches (e.g. 10X reaction), additive background (contaminating RNA), multiplicative platform effect between scRNA technologies.

3. Cell2location model for mapping to Nanostring WTA data (LocationModelWTA). See https://github.com/vitkl/SpaceJam for a new more versatile version.

4. Similified versions of model #1 that lack particular features of the full model, accessible from cell2location.models.simplified

Additionally we provide 2 models for downstream analysis of cell abundance estimates, accessible from cell2location.models.downstream:

1. CoLocatedGroupsSklearnNMF - identifying groups of cell types with similar locations using NMF (wrapper around sklearn NMF). See tutorial for usage.

2. ArchetypalAnalysis - identifying smoothly varying and mutually exclusive tissue zones with Archetypa Analysis.

Common errors, using containers:

• Common errors
  • Pyro/scvi-tools version
  • Pymc3 version (advanced use)
• Using docker image
• Using singularity image

API:

• Initial gene filtering
  • Show and apply gene filtering
• Reference signatures (NB regression)
  • User-facing reference signature estimation model class (scvi-tools BaseModelClass)
  • Pyro and scvi-tools Module classes
    • Pyro Module class (defining the model using pyro)
    • scvi-tools Module class (initialising the model and the guide, PyroBaseModuleClass)
• Reference signatures (hard-coded, cell type hierarchy)
  • Compute per cluster average expression (average RNA counts)
  • Decompose reference signatures according to levels of cell type hierarchy
  • Select genes based on DE (experimental)
• Cell2location: spatial mapping (scvi-tools/pyro)
  • User-facing cell2location spatial cell abundance estimation model class (scvi-tools BaseModelClass)
  • Pyro and scvi-tools Module classes (inc math description)
    • Pyro Module class (defining the model using pyro, math description)
  • Simplified model architectures
    • No prior factorisation of w_sf (Pyro Module class, math description)
    • No gene-specific platform effect m_g (Pyro Module class, math description)
• Downstream Analysis Models
  • Identifying cellular compartments / tissue zones (sklearn NMF)
  • Archetypal Analysis
• Plotting
  • Plot cell abundance for multiple cell types using colour interpolation
  • Plot heatmap and dotplot (hierarchically clustered but without showing the tree)
  • Other plotting functions
    • Rotate and crop image and location coordinates
    • Video illustrating mapping
    • Expected vs Observed model result QC plot
    • Plot factor spatial (legacy, now using scanpy.pl.spatial)
  • Module contents
• General utils
  • Select one Visium slide from anndata object
  • Compute KNN graph using spatial coordinates
  • Add the values of neighbours in a KNN graph for each index node
• Pyro and scvi-tools infrastructure classes
  • Base mixin classes (AutoGuide setup, posterior quantile computation, plotting & export)
  • scvi-tools Module classes (initialising the model and the guide, PyroBaseModuleClass)
    • Cell2location spatial cell abundance estimation
    • Reference signature estimation
• Statistical distribution classes
  • AutoNormalEncoder class for automatic generation of amortised Variational approximation
  • Other module contents

Nanostring WTA Tutorial (pymc3):

• Installation of pymc3 version with GPU support
• Running cell2location on NanostringWTA data

Indices and tables

• Module Index

• Search Page