AnnData structure and usage

The AnnData object stores a data matrix X together with annotations of observations obs (obsm, obsp) and variables var (varm, varp). Additional layers of data can be stored in layers and unstructured annotations in uns.

Back ends

There are different back ends for AnnData objects that inherit from the abstract AbstractAnnData class. For example, the InMemoryAnnData stores data in memory or the HDF5AnnData is backed by a H5AD file.

Usage

The items listed as "Slots" are elements of the AnnData object that contain data and can be accessed or set. "Fields" return information about the AnnData object but cannot be set directly. Both, as well as methods, can be accessed using the $ operator

For example:

adata$X returns the X matrix
adata$X <- x sets the X matrix
adata$method() calls a method

Fields

shape: Dimensions (observations x variables) of the AnnData object
n_obs: Number of observations
n_vars: Number of variables
obs_keys: Keys (column names) of obs
var_keys: Keys (column names) of var
layers_keys: Keys (element names) of layers
obsm_keys: Keys (element names) of obsm
varm_keys: Keys (element names) of varm
obsp_keys: Keys (element names) of obsp
varp_keys: Keys (element names) of varp
uns_keys: Keys (element names) of uns

Slots

X: The main data matrix. Either NULL or an observation x variable matrix (without dimnames) with dimensions consistent with n_obs and n_vars.
layers: Additional data layers. Must be NULL or a named list of matrices having dimensions consistent with n_obs and n_vars.
obs: Observation annotations. A data.frame with columns containing information about observations. The number of rows of obs defines the observation dimension of the AnnData object (n_obs). If NULL, an n_obs × 0 data.frame will automatically be generated.
var: Variable annotations. A data.frame with columns containing information about variables. The number of rows of var defines the variable dimension of the AnnData object (n_vars). If NULL, an n_vars × 0 data.frame will automatically be generated.
obs_names: Observation names. Either NULL or a vector of unique identifiers used to identify each row of obs and to act as an index into the observation dimension of the AnnData object. For compatibility with R representations, obs_names should be a unique character vector.
var_names: Variable names. Either NULL or a vector of unique identifiers used to identify each row of var and to act as an index into the variable dimension of the AnnData object. For compatibility with R representations, var_names should be a unique character vector.
obsm: Multi-dimensional observation annotation. Must be NULL or a named list of array-like elements with number of rows equal to n_obs.
varm: Multi-dimensional variable annotations. Must be NULL or a named list of array-like elements with number of rows equal to n_vars.
obsp: Observation pairs. Must be NULL or a named list of array-like elements with number of rows and columns equal to n_obs.
varp: Variable pairs. Must be NULL or a named list of array-like elements with number of rows and columns equal to n_vars.
uns: Unstructured annotations. Must be NULL or a named list.

Methods

Conversion methods:

as_SingleCellExperiment(): Convert to SingleCellExperiment::SingleCellExperiment, see as_SingleCellExperiment()
as_Seurat(): Convert to SeuratObject::Seurat, see as_Seurat()
as_InMemoryAnnData(): Convert to InMemoryAnnData, as as_InMemoryAnnData()
as_HDF5AnnData(): Convert to HDF5AnnData, see as_HDF5AnnData()

Output methods:

write_h5ad(): Write the AnnData object to an HDF5 file, see write_h5ad()

General methods:

print(): Print a summary of the AnnData object

Functions that can be used to create AnnData objects

AnnData(): Create an InMemoryAnnData object
read_h5ad(): Read an AnnData from a H5AD file
as_AnnData(): Convert other objects to an AnnData object