Quick start
This page provides details to start using the Census. Click here for more detailed Python tutorials (R vignettes coming soon).
Contents:
Installation
Install the Census API by following these instructions.
Python quick start
Below are 3 examples of common operations you can do with the Census. As a reminder, the reference documentation for the API can be accessed via help():
import cellxgene_census
help(cellxgene_census)
help(cellxgene_census.get_anndata)
# etc
Querying a slice of cell metadata
The following reads the cell metadata and filters female cells of cell type microglial cell or neuron, and selects the columns assay, cell_type, tissue, tissue_general, suspension_type, and disease.
import cellxgene_census
with cellxgene_census.open_soma() as census:
# Reads SOMADataFrame as a slice
cell_metadata = census["census_data"]["homo_sapiens"].obs.read(
value_filter = "sex == 'female' and cell_type in ['microglial cell', 'neuron']",
column_names = ["assay", "cell_type", "tissue", "tissue_general", "suspension_type", "disease"]
)
# Concatenates results to pyarrow.Table
cell_metadata = cell_metadata.concat()
# Converts to pandas.DataFrame
cell_metadata = cell_metadata.to_pandas()
print(cell_metadata)
The output is a pandas.DataFrame with over 300K cells meeting our query criteria and the selected columns.
The "stable" release is currently 2023-07-25. Specify 'census_version="2023-07-25"' in future calls to open_soma() to ensure data consistency.
assay cell_type tissue tissue_general suspension_type disease sex
0 10x 3' v3 microglial cell eye eye cell normal female
1 10x 3' v3 microglial cell eye eye cell normal female
2 10x 3' v3 microglial cell eye eye cell normal female
3 10x 3' v3 microglial cell eye eye cell normal female
4 10x 3' v3 microglial cell eye eye cell normal female
... ... ... ... ... ... ... ...
379219 microwell-seq neuron adrenal gland adrenal gland cell normal female
379220 microwell-seq neuron adrenal gland adrenal gland cell normal female
379221 microwell-seq neuron adrenal gland adrenal gland cell normal female
379222 microwell-seq neuron adrenal gland adrenal gland cell normal female
379223 microwell-seq neuron adrenal gland adrenal gland cell normal female
[379224 rows x 7 columns]
Obtaining a slice as AnnData
The following creates an anndata.AnnData object on-demand with the same cell filtering criteria as above and filtering only the genes ENSG00000161798, ENSG00000188229.
import cellxgene_census
with cellxgene_census.open_soma() as census:
adata = cellxgene_census.get_anndata(
census = census,
organism = "Homo sapiens",
var_value_filter = "feature_id in ['ENSG00000161798', 'ENSG00000188229']",
obs_value_filter = "sex == 'female' and cell_type in ['microglial cell', 'neuron']",
column_names = {"obs": ["assay", "cell_type", "tissue", "tissue_general", "suspension_type", "disease"]},
)
print(adata)
The output with about 300K cells and 2 genes can be now used for downstream analysis using scanpy.
AnnData object with n_obs × n_vars = 379224 × 2
obs: 'assay', 'cell_type', 'tissue', 'tissue_general', 'suspension_type', 'disease', 'sex'
var: 'soma_joinid', 'feature_id', 'feature_name', 'feature_length'
Memory-efficient queries
This example provides a demonstration to access the data for larger-than-memory operations using TileDB-SOMA operations.
First we initiate a lazy-evaluation query to access all brain and male cells from human. This query needs to be closed — query.close() — or called in a context manager — with ....
import cellxgene_census
import tiledbsoma
with cellxgene_census.open_soma() as census:
human = census["census_data"]["homo_sapiens"]
query = human.axis_query(
measurement_name = "RNA",
obs_query = tiledbsoma.AxisQuery(
value_filter = "tissue == 'brain' and sex == 'male'"
)
)
# Continued below
Now we can iterate over the matrix count, as well as the cell and gene metadata. For example, to iterate over the matrix count, we can get an iterator and perform operations for each iteration.
# Continued from above
iterator = query.X("raw").tables()
# Get an iterative slice as pyarrow.Table
raw_slice = next (iterator)
...
And you can now perform operations on each iteration slice. As with any any Python iterator this logic can be wrapped around a for loop.
And you must close the query.
# Continued from above
query.close()
R quick start
Below are 3 examples of common operations you can do with the Census. As a reminder, the reference documentation for the API can be accessed via ?:
library("cellxgene.census")
?cellxgene.census::get_seurat
Querying a slice of cell metadata
The following reads the cell metadata and filters female cells of cell type microglial cell or neuron, and selects the columns assay, cell_type, tissue, tissue_general, suspension_type, and disease.
The cellxgene.census package uses R6 classes and we recommend you to get familiar with their usage.
library("cellxgene.census")
census <- open_soma()
# Open obs SOMADataFrame
cell_metadata <- census$get("census_data")$get("homo_sapiens")$get("obs")
# Read as Arrow Table
cell_metadata <- cell_metadata$read(
value_filter = "sex == 'female' & cell_type %in% c('microglial cell', 'neuron')",
column_names = c("assay", "cell_type", "sex", "tissue", "tissue_general", "suspension_type", "disease")
)
# Concatenates results to an Arrow Table
cell_metadata <- cell_metadata$concat()
# Convert to R tibble (dataframe)
cell_metadata <- as.data.frame(cell_metadata)
print(cell_metadata)
census$close()
The output is a tibble with over 300K cells meeting our query criteria and the selected columns.
# A tibble: 379,224 × 7
assay cell_type sex tissue tissue_general suspension_type disease
<chr> <chr> <chr> <chr> <chr> <chr> <chr>
1 10x 3' v3 microglial cell fema… eye eye cell normal
2 10x 3' v3 microglial cell fema… eye eye cell normal
3 10x 3' v3 microglial cell fema… eye eye cell normal
4 10x 3' v3 microglial cell fema… eye eye cell normal
5 10x 3' v3 microglial cell fema… eye eye cell normal
6 10x 3' v3 microglial cell fema… eye eye cell normal
7 10x 3' v3 microglial cell fema… eye eye cell normal
8 10x 3' v3 microglial cell fema… eye eye cell normal
9 10x 3' v3 microglial cell fema… eye eye cell normal
10 10x 3' v3 microglial cell fema… eye eye cell normal
# ℹ 379,214 more rows
# ℹ Use `print(n = ...)` to see more rows
Obtaining a slice as a Seurat or SingleCellExperiment object
The following creates a Seurat object on-demand with a smaller set of cells and filtering only the genes ENSG00000161798, ENSG00000188229.
library("cellxgene.census")
library("Seurat")
census <- open_soma()
organism <- "Homo sapiens"
gene_filter <- "feature_id %in% c('ENSG00000107317', 'ENSG00000106034')"
cell_filter <- "cell_type == 'sympathetic neuron'"
cell_columns <- c("assay", "cell_type", "tissue", "tissue_general", "suspension_type", "disease")
seurat_obj <- get_seurat(
census = census,
organism = organism,
var_value_filter = gene_filter,
obs_value_filter = cell_filter,
obs_column_names = cell_columns
)
print(seurat_obj)
The output with over 4K cells and 2 genes can be now used for downstream analysis using Seurat.
An object of class Seurat
2 features across 4744 samples within 1 assay
Active assay: RNA (2 features, 0 variable features)
Similarly a SingleCellExperiment object can be created.
library("SingleCellExperiment")
sce_obj <- get_single_cell_experiment(
census = census,
organism = organism,
var_value_filter = gene_filter,
obs_value_filter = cell_filter,
obs_column_names = cell_columns
)
print(sce_obj)
The output with over 4K cells and 2 genes can be now used for downstream analysis using the Bioconductor ecosystem.
class: SingleCellExperiment
dim: 2 4744
metadata(0):
assays(1): counts
rownames(2): ENSG00000106034 ENSG00000107317
rowData names(2): feature_name feature_length
colnames(4744): obs48350835 obs48351829 ... obs52469564 obs52470190
colData names(6): assay cell_type ... suspension_type disease
reducedDimNames(0):
mainExpName: RNA
altExpNames(0):
Memory-efficient queries
This example provides a demonstration to access the data for larger-than-memory operations using TileDB-SOMA operations.
First we initiate a lazy-evaluation query to access all brain and male cells from human. This query needs to be closed — query$close().
library("cellxgene.census")
library("tiledbsoma")
human <- census$get("census_data")$get("homo_sapiens")
query <- human$axis_query(
measurement_name = "RNA",
obs_query = SOMAAxisQuery$new(
value_filter = "tissue == 'brain' & sex == 'male'"
)
)
# Continued below
Now we can iterate over the matrix count, as well as the cell and gene metadata. For example, to iterate over the matrix count, we can get an iterator and perform operations for each iteration.
# Continued from above
iterator <- query$X("raw")$tables()
# For sparse matrices use query$X("raw")$sparse_matrix()
# Get an iterative slice as an Arrow Table
raw_slice <- iterator$read_next()
#...
And you can now perform operations on each iteration slice. This logic can be wrapped around a while() loop and checking the iteration state by monitoring the logical output of iterator$read_complete()
And you must close the query and census.
# Continued from above
query.close()
census.close()