In waldronlab/SingleCellMultiModal: Integrating Multi-modal Single Cell Experiment datasets

Installation

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")

BiocManager::install("SingleCellMultiModal")

Load libraries

library(MultiAssayExperiment)
library(SingleCellMultiModal)
library(SingleCellExperiment)

ECCITE-seq dataset

ECCITE-seq data are an evolution of the CITE-seq data (see also CITE-seq vignette for more details) by extending the CITE-seq original data types with a third one always extracted from the same cell. Indeed, in addition to the CITE-seq providing scRNA-seq and antibody-derived tags (ADT), it provides around ten Hashtagged Oligo (HTO). In particular this dataset is provided by @mimitou2019multiplexed.

Downloading datasets

The user can see the available dataset by using the default options through the CITE-seq function.

CITEseq(DataType="peripheral_blood", modes="*", dry.run=TRUE, version="1.0.0")

Or simply by setting dry.run = FALSE it downloads the data and by default creates the MultiAssayExperiment object.

In this example, we will use one of the two available datasets scADT_Counts:

mae <- CITEseq(DataType="peripheral_blood", modes="*", dry.run=FALSE, version="1.0.0")
mae

Example with actual data:

experiments(mae)

Additionally, we stored into the object metedata

Exploring the data structure

Check row annotations:

rownames(mae)

Take a peek at the sampleMap:

sampleMap(mae)

scRNA-seq data

The scRNA-seq data are accessible with the name scRNAseq, which returns a matrix object.

head(experiments(mae)$scRNA)[, 1:4]

scADT data

The scADT data are accessible with the name scADT, which returns a matrix object.

head(experiments(mae)$scADT)[, 1:4]

CTCL/CTRL conditions

The dataset has two different conditions (CTCL and CTRL) which samples can be identified with the colData accessor.

CTCL stands for cutaneous T-cell lymphoma while CTRL for control.

For example, if we want only the CTCL samples, we can run:

(ctclMae <- mae[,colData(mae)$condition == "CTCL",])

And if you're interested into the common samples across all the modalities you can use the complete.cases funtion.

ctclMae[,complete.cases(ctclMae),]

sgRNAs CRISPR pertubation data

The CRISPR perturbed scRNAs data are stored in a different spot to keep their original long format.

They can be accessed with the metadata accessors which, in this case returns a named list of data.frames.

sgRNAs <- metadata(mae)
names(sgRNAs)

There are four different sgRNAs datasets, one per each condition and family receptors combination.

TCR stands for T-Cell Receptor, while a,b,g,d stand for alpha, beta, gamma and delta respectively.

To look into the TCRab, simply run:

head(sgRNAs$CTCL_TCRab)

SingleCellExperiment object conversion

Because of already large use of some methodologies (such as in the [SingleCellExperiment vignette][1] or [CiteFuse Vignette][2] where the SingleCellExperiment object is used for CITE-seq data, we provide a function for the conversion of our CITE-seq MultiAssayExperiment object into a SingleCellExperiment object with scRNA-seq data as counts and scADT data as altExps.

sce <- CITEseq(DataType="peripheral_blood", modes="*", dry.run=FALSE, 
               version="1.0.0", DataClass="SingleCellExperiment")
sce

Session Info

sessionInfo()

Additional References

https://www.bioconductor.org/packages/release/bioc/vignettes/SingleCellExperiment/inst/doc/intro.html#5_adding_alternative_feature_sets http://www.bioconductor.org/packages/release/bioc/vignettes/CiteFuse/inst/doc/CiteFuse.html

References

waldronlab/SingleCellMultiModal documentation built on Nov. 3, 2024, 7:32 p.m.