Data processing

get_celltypes(a::AnnData)

Tries to infer the cell types of cells in an AnnData object.

Returns a vector of cell type names if the cell types are stored in adata.obs["cell_type"], adata.obs["celltype"], adata.obs["celltypes"], or adata.obs["cell_types"]. Otherwise, returns nothing.

subset_adata(adata::AnnData, subset_inds::Tuple, dims::Symbol=:both)

Subset an AnnData object by indices passed as a tuple of vectors of integers, UnitRanges, or vectors of Booleans. If dims is set to :both, the first element of subset_inds is used to subset cells and the second element is used to subset genes.

Arguments

• adata: AnnData object to subset
• subset_inds: tuple of vectors of integers, UnitRanges, or vectors of Booleans
• dims: dimension to subset, either :cells, :genes, or :both

Returns

• a copy of the AnnData object with the subsetted data

subset_adata(adata::AnnData, subset_inds::Union{Int, Vector{Int}, UnitRange, Vector{Bool}}, dims::Symbol)

Subset an AnnData object by indices passed as a vector of integers or booleans or as a UnitRange. The dims argument can be set to either :cells or :genes to specify which dimension to subset.

Arguments

• adata: AnnData object to subset
• subset_inds: vector of integers or booleans or UnitRange
• dims: dimension to subset, either :cells or :genes

Returns

• a copy of the AnnData object with the subsetted data

subset_adata!(adata::AnnData, subset_inds, dims::Symbol)

In-place version of subset_adata, see ?subset_adata for more details.

For subset_inds, either a tuple of vectors or ranges can be passed with dims set to :both, for subsetting both cells and genes, or a single vector or range can be passed with dims set to either :cells or :genes.

Arguments

• adata: AnnData object to subset
• subset_inds: tuple of vectors of integers, UnitRanges, or vectors of Booleans or vector of integers or booleans or UnitRange
• dims: dimension to subset, either :cells, :genes, or :both

Returns

• the AnnData object with the subsetted data

init_library_size(adata::AnnData; batch_key::Symbol=:batch)

Computes and returns library size based on AnnData object.

Based on the scvi-tools function from here

Returns a tupe of arrays of length equal to the number of batches in adata as stored in adata.obs[!,:batch_key], containing the means and variances of the library size in each batch in adata. Default batch key: :batch, if it is not found, defaults to 1 batch.

Arguments

• adata: AnnData object

Keyword arguments

• batch_key: the key in adata.obs that stores the batch information (default: :batch)

Returns

• a tuple of arrays of length equal to the number of batches in adata as stored in adata.obs[!,:batch_key], corresponding to the log library means and vars per batch

estimate_size_factors(mat; locfunc=median)

Estimates size factors to use for normalization, based on the corresponding Seurat functionality. Assumes a countmatrix mat in cell x gene format as input, returns a vector of size factors.

For details, please see the Seurat documentation.

Arguments

• mat: countmatrix in cell x gene format

Keyword arguments

• locfunc: the function to use for calculating the location parameter (default: median)

Returns

• a vector of size factors

normalize_size_factors(mat::Abstractmatrix)

Normalizes the countdata in mat by dividing it by the size factors calculated with estimate_size_factors. Assumes a countmatrix mat in cell x gene format as input, returns the normalized matrix.

Arguments

• mat: countmatrix in cell x gene format

Returns

• the normalized countmatrix

normalize_size_factors(adata::AnnData)

Normalizes the adata.X by dividing it by the size factors calculated with estimate_size_factors. Adds the normalized count matrix to adata.layers and returns adata.

Arguments

• adata: AnnData object

Returns

• the modified AnnData object

normalize_total!(adata::AnnData; 
    target_sum::Union{Nothing, Real}=nothing, 
    key_added::String="cell_counts",
    layer::Union{Nothing, String}=nothing,
    verbose::Bool=false)

Normalizes counts per cell, specifically normalize each cell by total counts over all genes, so that every cell has the same total count after normalization. If choosing target_sum=1e6, this is CPM normalization.

Basic version of the scanpy.pp.normalize_total function

Arguments

• adata: AnnData object

Keyword arguments

• target_sum: if nothing, after normalization, each observation (cell) has a total count equal to the median of total counts for observations (cells) before normalization.
• key_added: name of the field in adata.obs where the normalization factor is stored, set to "cell_counts" by default
• layer: optional; which layer to normalize on. If nothing, adata.X is used.

Returns

• the adata object with normalized version of the original adata.X in adata.layers and the size factors in adata.obs.

normalize_total(adata::AnnData; 
    target_sum::Union{Nothing, Real}=1e4, 
    key_added::String="cell_counts",
    verbose::Bool=false)

Normalizes counts per cell, specifically normalize each cell by total counts over all genes, so that every cell has the same total count after normalization. See normalize_total! for details. Unlike the in-place version, this function returns a dictionary with the normalized counts and scaled counts per cell.

Arguments

• adata: AnnData object

Keyword arguments

• target_sum: if nothing, after normalization, each observation (cell) has a total count equal to the median of total counts for observations (cells) before normalization.
• key_added: name of the field in adata.obs where the normalization factor is stored, set to "cell_counts" by default
• verbose: whether to print progress messages (default: false)

Returns

• a dictionary with the normalized counts and scaled counts per cell

function filter_cells(countmatrix::AbstractMatrix; 
    min_counts::Union{Int, Nothing}=nothing, 
    min_genes::Union{Int, Nothing}=nothing, 
    max_counts::Union{Int, Nothing}=nothing, 
    max_genes::Union{Int, Nothing}=nothing,
    verbose::Bool=true,
    make_plot::Bool=false,
    log_transform_plot::Bool=false,
    save_plot::Bool=false,
    plot_filename::String="gene_filtering_histogram.pdf"
    )

Filter cell outliers based on counts and numbers of genes expressed. For instance, only keep cells with at least min_counts counts or min_genes genes expressed. This is to filter measurement outliers, i.e. “unreliable” observations. Only provide one of the optional parameters min_counts, min_genes, max_counts, max_genes per call.

Arguments

• countmatrix: countmatrix of shape n_obs × n_vars. Rows correspond to cells and columns to genes.
• min_counts: Minimum number of counts required for a cell to pass filtering.
• min_genes: Minimum number of genes expressed required for a cell to pass filtering.
• max_counts: Maximum number of counts required for a cell to pass filtering.
• max_genes: Maximum number of genes expressed required for a cell to pass filtering.

Keyword arguments

• verbose: whether to print info and status messages
• make_plot: whether to plot the histogram of cells before filtering
• log_transform_plot: whether to log-transform the counts of cells for plotting
• save_plot: whether to save the histogram of cells before filtering
• plot_filename: filename to save the histogram of cells before filtering

Returns

• cells_subset: BitVector index mask that does filtering; true means that the cell is kept, false means the cell is removed.
• number_per_cell: Depending on what was thresholded (counts or genes), the array stores n_counts or n_genes per cell.

function filter_cells!(adata::AnnData; 
    min_counts::Union{Int, Nothing}=nothing, 
    min_genes::Union{Int, Nothing}=nothing, 
    max_counts::Union{Int, Nothing}=nothing, 
    max_genes::Union{Int, Nothing}=nothing,
    verbose::Bool = true, 
    plot_before::Bool=false,
    log_transform_plot::Bool=false,
    plot_after::Bool=false,
    save_plot_before::Bool=false, 
    save_plot_after::Bool=false,
    plot_filename_before::String="cell_histogram_before_filtering.pdf",
    plot_filename_after::String="cell_histogram_after_filtering.pdf"
    )

Filter cell outliers based on counts and numbers of genes expressed. For instance, only keep cells with at least min_counts counts or min_genes genes expressed. This is to filter measurement outliers, i.e. “unreliable” observations. Only provide one of the optional parameters min_counts, min_genes, max_counts, max_genes per call.

Arguments

• adata: AnnData object of shape n_obs × n_vars. Rows correspond to cells and columns to genes.
• min_counts: Minimum number of counts required for a cell to pass filtering.
• min_genes: Minimum number of genes expressed required for a cell to pass filtering.
• max_counts: Maximum number of counts required for a cell to pass filtering.
• max_genes: Maximum number of genes expressed required for a cell to pass filtering

Keyword arguments

• verbose: whether to print info and status messages
• plot_before: whether to plot the histogram of cells before filtering
• log_transform_plot: whether to log-transform the counts of cells for plotting
• plot_after: whether to plot the histogram of cells after filtering
• save_plot_before: whether to save the histogram of cells before filtering
• save_plot_after: whether to save the histogram of cells after filtering
• plot_filename_before: filename to save the histogram of cells before filtering
• plot_filename_after: filename to save the histogram of cells after filtering

Returns

• the filtered AnnData object

filter_genes!(adata::AnnData; 
    min_counts::Union{Int, Nothing}=nothing, 
    min_cells::Union{Int, Nothing}=nothing, 
    max_counts::Union{Int, Nothing}=nothing, 
    max_cells::Union{Int, Nothing}=nothing,
    verbose::Bool = true,
    plot_before::Bool=false,
    log_transform_plot::Bool=false,
    plot_after::Bool=false,
    save_plot_before::Bool=false, 
    save_plot_after::Bool=false,
    plot_filename_before::String="gene_histogram_before_filtering.pdf",
    plot_filename_after::String="gene_histogram_after_filtering.pdf"
)

Filter genes based on number of cells or counts. Keep genes that have at least min_counts counts or are expressed in at least min_cells cells or have at most max_counts counts or are expressed in at most max_cells cells. Only provide one of the optional parameters min_counts, min_cells, max_counts, max_cells per call.

Arguments

• adata: AnnData object of shape n_obs × n_vars. Rows correspond

to cells and columns to genes.

• min_counts: Minimum number of counts required for a gene to pass filtering.
• min_cells: Minimum number of cells expressed required for a gene to pass filtering.
• max_counts: Maximum number of counts required for a gene to pass filtering.
• max_cells: Maximum number of cells expressed required for a gene to pass filtering.

Keyword arguments

• verbose: whether to print info and status messages
• plot_before: whether to plot the histogram of genes before filtering
• log_transform_plot: whether to log-transform the counts of genes for plotting
• plot_after: whether to plot the histogram of genes after filtering
• save_plot_before: whether to save the histogram of genes before filtering
• save_plot_after: whether to save the histogram of genes after filtering
• plot_filename_before: filename to save the histogram of genes before filtering
• plot_filename_after: filename to save the histogram of genes after filtering

Returns

• the filtered AnnData object

filter_genes(adata::AnnData; 
    min_counts::Union{Int, Nothing}=nothing, 
    min_cells::Union{Int, Nothing}=nothing, 
    max_counts::Union{Int, Nothing}=nothing, 
    max_cells::Union{Int, Nothing}=nothing,
    verbose::Bool=true,
    make_plot::Bool=false,
    log_transform_plot::Bool=false,
    save_plot::Bool=false,
    plot_filename::String="gene_filtering_histogram.pdf"
    )

Filter genes based on number of cells or counts. Keep genes that have at least min_counts counts or are expressed in at least min_cells cells or have at most max_counts counts or are expressed in at most max_cells cells. Only provide one of the optional parameters min_counts, min_cells, max_counts, max_cells per call. This is the out-of-place version; for details on the input arguments, see the in-place version filter_genes!

Arguments

• adata: AnnData object of shape n_obs × n_vars. Rows correspond to cells and columns to genes.
• min_counts: Minimum number of counts required for a gene to pass filtering.
• min_cells: Minimum number of cells in which it is expressed required for a gene to pass filtering.
• max_counts: Maximum number of counts required for a gene to pass filtering.
• max_cells: Maximum number of cells in which it is expressed required for a gene to pass filtering.

Keyword arguments

• verbose: whether to print info and status messages
• make_plot: whether to plot the histogram of cells before filtering
• log_transform_plot: whether to log-transform the counts of cells for plotting
• save_plot: whether to save the histogram of cells before filtering
• plot_filename: filename to save the histogram of cells before filtering

Returns

• gene_subset: BitVector index mask that does filtering; true means that the gene is kept, false means the cell is removed.
• number_per_gene: Depending on what was thresholded (counts or cells), the array stores n_counts or n_cells per gene.

log_transform!(adata::AnnData; 
    layer::String="normalized",
    verbose::Bool=false)

Log-transforms the data. Looks for a layer of normalized counts in adata.layers["normalized"]. If the layer is not there, it uses adata.X.

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for PCA (default: "log_transformed")
• verbose: whether to print progress messages (default: false)

Returns

• the adata object with the log-transformed values in a new layer "log_transformed".

logp1_transform!(adata::AnnData; 
    layer::Union{String, Nothing}=nothing,
    verbose::Bool=false)

Log-transforms the (count) data, adding a pseudocount of 1. Uses the X in adata.X by default, but other layers can be passed using the layer keyword.

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for log + 1 transformation
• verbose: whether to print progress messages (default: false)

Returns

• the adata object with the log-transformed values in a new layer "logp1_transformed".

sqrt_transform!(adata::AnnData; 
    layer::String="normalized",
    verbose::Bool=false)

Sqrt-transforms the data. Looks for a layer of normalized counts in adata.layers["normalized"]. If the layer is not there, it uses adata.X.

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for the transformation
• verbose: whether to print progress messages (default: false)

Returns

• the adata object with the sqrt-transformed values in a new layer "sqrt_transformed".

rescale!(adata::AnnData; 
    layer::Union{String, Nothing}=nothing,
    verbose::Bool=false)

Rescales the data to zero mean and unit variance in each gene, using the specified layer. If none is provided, it uses adata.X.

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for the transformation
• verbose: whether to print progress messages (default: false)

Returns

• the adata object with the rescales values in a new layer "rescaled".

function pca!(adata::AnnData; 
    layer::String="log_transformed", 
    n_pcs::Int=size(adata.X,2),
    verbose::Bool=false
)

Performs a PCA on the specified layer of an AnnData object and stores the results in adata.obsm. Uses all variables of the layer by default, but the number of PCs to be stored can also be specified with the n_pcs keyword. If the layer is not found, the log-transformed normalized counts are calculated and used.

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for PCA (default: "log_transformed")
• n_pcs: the number of PCs to be stored (default: all variables)
• verbose: whether to print progress messages (default: false)

Returns

• the modified AnnData object

function umap!(adata::AnnData; 
    layer::String="log_transformed", 
    use_pca_init::Bool=false, 
    n_pcs::Int=100, 
    verbose::Bool=true, 
    kwargs...)

Performs UMAP on the specified layer of an AnnData object. If the layer is not found, the log-transformed normalized counts are calculated and used. Optionally, UMAP can be run on a PCA representation, the number of PCs can be specified (default=100). For customizing the behaviour or UMAP, see the keyword arguments of the UMAP.UMAP_ function. They can all be passed via the kwargs.

The fields of the resulting UMAP_ struct are stored as follows: - the UMAP embedding in adata.obsm["umap"], - the fuzzy simplicial knn graph in adata.obsp["fuzzyneighborgraph"], - the KNNs of each cell in adata.obsm["knns"], - the distances of each cell to its KNNs in adata.obsm["knn_dists"]

Arguments

• adata: the AnnData object to be modified

Keyword arguments

• layer: the layer to be used for UMAP (default: "log_transformed")
• use_pca_init: whether to use a PCA representation for UMAP (default: false)
• n_pcs: the number of PCs to be used for UMAP (default: 100)
• verbose: whether to print progress messages (default: false)
• kwargs: keyword arguments for UMAP.UMAP_

Returns

• the modified AnnData object

highly_variable_genes(adata::AnnData;
    layer::Union{String,Nothing} = nothing,
    n_top_genes::Int=2000,
    batch_key::Union{String,Nothing} = nothing,
    span::Float64=0.3
    )

Computes highly variable genes according to the workflows on scanpy and Seurat v3 per batch and returns a dictionary with the information on the joint HVGs. For the in-place version, see highly_variable_genes!

More specifically, it is the Julia re-implementation of the corresponding scanpy function For implementation details, please check the scanpy/Seurat documentations or the source code of the lower-level _highly_variable_genes_seurat_v3 function in this package. Results are almost identical to the scanpy function. The differences have been traced back to differences in the local regression for the mean-variance relationship implemented in the Loess.jl package, that differs slightly from the corresponding Python implementation.

Arguments

• adata: AnnData object

Keyword arguments

• layer: optional; which layer to use for calculating the HVGs. Function assumes this is a layer of counts. If layer is not provided, adata.X is used.
• n_top_genes: optional; desired number of highly variable genes. Default: 2000.
• batch_key: optional; key where to look for the batch indices in adata.obs. If not provided, data is treated as one batch.
• span: span to use in the loess fit for the mean-variance local regression. See the Loess.jl docs for details.
• replace_hvgs: whether or not to replace the hvg information if there are already hvgs calculated. If false, the new values are added with a "_1" suffix. Default:true,
• verbose: whether or not to print info on current status

Returns

• a dictionary containing information on the highly variable genes, specifically containing the following keys is added:
  • highly_variable: vector of Bools indicating which genes are highly variable
  • highly_variable_rank: rank of the highly variable genes according to (corrected) variance
  • means: vector with means of each gene
  • variances: vector with variances of each gene
  • variances_norm: normalized variances of each gene
  • highly_variable_nbatches: if there are batches in the dataset, logs the number of batches in which each highly variable gene was actually detected as highly variable.

highly_variable_genes!(adata::AnnData;
    layer::Union{String,Nothing} = nothing,
    n_top_genes::Int=2000,
    batch_key::Union{String,Nothing} = nothing,
    span::Float64=0.3,
    replace_hvgs::Bool=true,
    verbose::Bool=false
    )

Computes highly variable genes per batch according to the workflows on scanpy and Seurat v3 in-place. This is the in-place version that adds an dictionary containing information on the highly variable genes directly to the adata.var and returns the modified AnnData object. For details, see the not-in-place version ?highly_variable_genes.

subset_to_hvg!(adata::AnnData;
    layer::Union{String,Nothing} = nothing,
    n_top_genes::Int=2000,
    batch_key::Union{String,Nothing} = nothing,
    span::Float64=0.3,
    verbose::Bool=true
)

Calculates highly variable genes with highly_variable_genes! and subsets the AnnData object to the calculated HVGs. For description of input arguments, see highly_variable_genes!

Arguments

• adata: AnnData object

Keyword arguments

• layer: optional; which layer to use for calculating the HVGs. Function assumes this is a layer of counts. If layer is not provided, adata.X is used.
• n_top_genes: optional; desired number of highly variable genes. Default: 2000.
• batch_key: optional; key where to look for the batch indices in adata.obs. If not provided, data is treated as one batch.
• span: span to use in the loess fit for the mean-variance local regression. See the Loess.jl docs for details.
• verbose: whether or not to print info on current status

Returns

• adata object subset to the calculated HVGs, both in the countmatrix/layer data used for HVG calculation and in the adata.var dictionary.

function plot_pca(
    adata::AnnData;
    color_by::String="",
    pcs::Vector{Int}=[1,2],
    recompute::Bool=false,
    save_plot::Bool=false,
    filename::String="PCA.pdf"
)

Plot a PCA embedding on a given AnnData object.

Arguments

• adata: AnnData object

Keyword arguments

• color_by: column name of adata.obs to color the plot by, or a gene name to color by expression. If neither is provided, celltypes will be used if present, otherwise all cells will be colored the same.
• pcs: which PCs to plot
• recompute: whether to recompute the PCA embedding or use an already existing one
• save_plot: whether to save the plot to a file
• filename: filename to save the plot to

Returns

• the plot object

function plot_umap(
    adata::AnnData;
    color_by::String="",
    recompute::Bool=false,
    save_plot::Bool=false,
    filename::String="UMAP.pdf"
    )

Plot a UMAP embedding on a given AnnData object.

Arguments

• adata: AnnData object

Keyword arguments

• color_by: column name of adata.obs to color the plot by, or a gene name to color by expression. If neither is provided, celltypes will be used if present, otherwise all cells will be colored the same.
• recompute: whether to recompute the UMAP embedding
• save_plot: whether to save the plot to a file
• filename: filename to save the plot to

Returns

• the plot object

function plot_histogram(adata::AnnData, 
    cell_gene::Symbol = :gene, 
    counts_number::Symbol = :counts; 
    cutoff::Union{Nothing, Real}=nothing,
    log_transform::Bool=false, 
    save_plot::Bool=false, 
    filename::String="counts_per_gene_histogram.pdf"
    )

Plot a histogram of counts per gene or cells, or alternatively he number of genes expressed per cell or the number of cells per gene in which it is expressed.

Additionally, a user-specified cutoff can be plotted on top, to be used as a visualization tool for filtering.

The function is called internally when using the plotting options in the filter_cells and filter_genes functions.

Arguments

• adata: AnnData object
• cell_gene: one of :cell or gene; whether to plot counts per gene or per cell
• counts_number: one of :counts or :number; whether to plot counts or number of genes/cells

Keyword arguments

• cutoff: cutoff to plot on top of the histogram
• log_transform: whether to log transform the counts
• save_plot: whether to save the plot to a file
• filename: filename to save the plot to

Returns

• hist: the plot object

plot_highest_expressed_genes(adata::AnnData; 
    n_top::Int=30, 
    gene_symbols::Union{String, Nothing}=nothing,
    save_plot::Bool=false,
    filename::String="highest_expressed_genes.pdf"
)

The function computes for each gene the fraction of counts assigned to that gene within a cell. The n_top genes with the highest mean fraction over all cells are plotted as boxplots.

Arguments

• adata: AnnData object

Keyword arguments

• n_top: number of genes to plot
• gene_symbols: column name of adata.var to use as gene names. If nothing, adata.var_names will be used.
• save_plot: whether to save the plot to a file
• filename: filename to save the plot to

Returns

• the plot object

function plot_highly_variable_genes(adata::AnnData;
    log_transform::Bool=false,
    save_plot::Bool=false,
    filename::String="highly_variable_genes.pdf"
    )
end

Plot dispersions or normalized variance versus means for genes

Arguments

• adata: AnnData object

Keyword arguments

• log_transform: whether to log transform the counts
• save_plot: whether to save the plot to a file
• filename: filename to save the plot to

Returns

• the plot object

load_cortex(path::String=""; verbose::Bool=false)

Loads cortex dataset from Zeisel et al. 2015 and creates a corresponding AnnData object.

Looks for a file cortex_anndata.h5ad that can be downloaded from this GoogleDrive data folder. The functions first looks in the folder passed as path (default: assumes files are in a subfolder named data of the current directory, i.e., that the complete GoogleDrive data folder has been downloaded in the current directory), and alternatively downloads the data if is cannot find the file in the given path (see below).

The file is the h5 export of the Python AnnData object provided as built-in cortex dataset from scvi-tools, data is from Zeisel et al. 2015.

If the file is present, the data is loaded from the Python AnnData object and stored in an analogous Julia AnnData object. This is handled by the functions init_cortex_from_h5ad and load_cortex_from_h5ad.

Alternatively, if the h5ad file is not found in the folder, the data is downloaded directly from the original authors and processed analogous to the scvi-tools processing, and subsequently stored to a Julia AnnData object. This is handled by the function init_cortex_from_url.

Arguments

• path::String: path to the folder where the file is stored
• verbose::Bool: whether to print progress information

Returns

• the Julia AnnData object.

Example

julia> load_cortex()
    AnnData object with a countmatrix with 3005 cells and 1200 genes
    layers dict with the following keys: ["counts"]
    summary statistics dict with the following keys: ["n_labels", "n_vars", "n_batch", "n_continuous_covs", "n_cells", "n_proteins"]
    unique celltypes: ["interneurons", "pyramidal SS", "pyramidal CA1", "oligodendrocytes", "microglia", "endothelial-mural", "astrocytes_ependymal"]
    training status: not trained

load_pbmc(path::String = "data/")

Loads pbmc dataset from Zheng et al. 2017 and creates a corresponding AnnData object. Specifically, the PBMC8k version is used, preprocessed according to the Bioconductor workflow.

Loads the following files that can be downloaded from this GoogleDrive data folder:

• PBMC_counts.csv: countmatrix
• PBMC_annotation.csv: cell type annotation

Files are loaded from the folder passed as path (default: assumes files are in a subfolder named data of the current directory, i.e., that the complete GoogleDrive data folder has been downloaded in the current directory.)

From these input files, a Julia AnnData object is created. The countmatrix contains information on cell barcodes and gene names. The gene name and celltype information is stored in the vars and obs dictionaries of the AnnData object, respectively.

Arguments

• path: path to the folder containing the input files (default: "data/")

Returns

• the Julia AnnData object.

Example

julia> load_pbmc()
    AnnData object with a countmatrix with 7480 cells and 200 genes
    unique celltypes: ["B-cells", "CD4+ T-cells", "Monocytes", "CD8+ T-cells", "NK cells", "NA", "HSC", "Erythrocytes"]
    training status: not trained

load_tasic(path::String = "data/")

Loads tasic dataset based on Tasic et al. (2016) and creates a corresponding AnnData object.

Loads the following files that can be downloaded from this GoogleDrive data folder:

• Tasic_countmat.txt: countmatrix
• Tasic_celltypes.txt: cell types
• Tasic_genenames.txt: gene names
• Tasic_receptorandmarkers.txt: List of receptor and marker genes

Files are loaded from the folder passed as path (default: assumes files are in a subfolder named data of the current directory, i.e., that the complete GoogleDrive data folder has been downloaded in the current directory.

The original data is available at Gene expression Omnibus (GEO) under accession number GSE71585. Preprocessing and annotation has been prepared according to the original manuscript.

From these input files, a Julia AnnData object is created. The list of receptor and marker genes is used to annotate cells as neural vs. non-neural, and annotate the neural cells as GABA- or Glutamatergic.

These annotations together with the cell type information and the gene names and receptor/marker list are stored in Dictionaries in the obs and vars fields of the AnnData object.

Additionally, size factors are calculated and used for normalizing the counts. The normalized counts are stored in an additional layer named normalized_counts.

Arguments

• path: path to the folder containing the input files (default: "data/")

Returns

• the Julia AnnData object.

Example

julia> load_tasic()
    AnnData object with a countmatrix with 1679 cells and 15119 genes
    layers dict with the following keys: ["normalized_counts", "counts"]
    unique celltypes: ["Vip", "L4", "L2/3", "L2", "Pvalb", "Ndnf", "L5a", "SMC", "Astro", "L5", "Micro", "Endo", "Sst", "L6b", "Sncg", "Igtp", "Oligo", "Smad3", "OPC", "L5b", "L6a"]
    training status: not trained

subset_tasic!(adata::AnnData)

Subsets an input AnnData object initialized from the Tasic data according load_tasic to the neural cells and the receptor and marker genes provided as annotation.

Specifically, the count matrix and the normalized count matrix are subset to these cells and genes, and the dictionaries with information about cells and genes in adata.obs and adata.vars are also subset accordingly.

Arguments

• adata: AnnData object initialized from the Tasic data

Returns

• the modified AnnData object.