In MarioniLab/MouseThymusAgeing: Single-cell Transcriptomics Data of the Ageing Mouse Thymus

library(BiocStyle)
knitr::opts_chunk$set(error=FALSE, message=FALSE, warning=FALSE)

Overview

The purpose of this notebook is to prepare the mouse thymus ageing single-cell transcriptome data generated using the plate-based SMART-seq2 chemistry. Specific thymic epithelial cell (TEC) subpopulations (mTEClo, mTEChi, cTEC & Dsg3+ mTEC) were flourescence-activated cell sorted (FACS) into 384-well plates containing lysis buffer, and frozen down before processing using the SMART-seq2 chemistry. Single-cells were sampled from mice of 5 different ages: 1, 4, 16, 32 and 52 weeks old. In addition, to minimise potential batch effect confounding, ages were mixed on plates such that each plate contained cells from 3 ages. The exact locations were different between plates to remove the potential for plate-spatial effects to also confound analyses downstream. Finally, the sorting took place over 5 days, with cells derived from separate mice on each day, leading to 5 replicate mice for each combination of time point and TEC subpopulation.

Preparing the processed data

To minimise processing and maximise usability, we have provided a single gzip compressed counts matrix after removal of poor-quality cells. Borrowing from the r Biocpkg("MouseGastrulationAtlas") we also make using of caching in the r Biocpkg("BiocFileCache") to avoid having to constantly download data for subsequent analyses.

library(BiocFileCache)
bfc <- BiocFileCache("SMARTseq_raw", ask=FALSE)
count.path <- bfcrpath(bfc, file.path("https://content.cruk.cam.ac.uk/",
                                      "jmlab/thymus_data/QC_counts.tsv.gz"))

These contain the gene counts for cells that pass QC (detailed in manuscript); I have not removed the genes with all zero counts across these cells. For these data to be usable we also need the gene IDs - these are provided as a separate table that contains additional information such as the chromosome, start and end position and gene length.

genes.path <- bfcrpath(bfc, file.path("https://content.cruk.cam.ac.uk/",
                                      "jmlab/thymus_data/Gene_info.tsv"))

Both of these data can be loaded in using the standard read.table functions.

thymus.counts <- read.table(count.path, sep="\t", header=TRUE, stringsAsFactors=FALSE)
gene.info <- read.table(genes.path, sep="\t", header=TRUE, stringsAsFactors=FALSE)

Finally, we can download the size factors and other cell-wise meta-data, including sorting day (replicate), sort population, age, cluster ID as assigned in the manuscript and plate information. We can also pull in the reduced dimensional representations at this stage.

# meta data
meta.path <- bfcrpath(bfc, file.path("https://content.cruk.cam.ac.uk/",
                                     "jmlab/thymus_data/meta_data.tsv.gz"))
meta.data <- read.table(meta.path, sep="\t", header=TRUE, stringsAsFactors=FALSE)
rownames(meta.data) <- meta.data$CellID
meta.data <- meta.data[colnames(thymus.counts), ]
head(meta.data)

# pca
reds.path <- bfcrpath(bfc, file.path("https://content.cruk.cam.ac.uk/",
                                     "jmlab/thymus_data/reduced_dims.tsv.gz"))
pca.dims <- read.table(reds.path, sep="\t", header=TRUE, stringsAsFactors=FALSE)
rownames(pca.dims) <- pca.dims$CellID
pca.dims <- pca.dims[colnames(thymus.counts), ]
head(pca.dims)

# size factors
sf.path <- bfcrpath(bfc, file.path("https://content.cruk.cam.ac.uk/",
                                     "jmlab/thymus_data/size_factors.tsv.gz"))
size.factors <- read.table(sf.path, sep="\t", header=TRUE, stringsAsFactors=FALSE)
rownames(size.factors) <- size.factors$CellID
size.factors <- size.factors[colnames(thymus.counts), ]
head(size.factors)

We will now put all of these together into a convenient SingleCellExperiment object.

library(SingleCellExperiment)
thymus.sce <- SingleCellExperiment(assays=list(counts=thymus.counts),
                                   colData=meta.data, rowData=gene.info)

# add the size factors
sizeFactors(thymus.sce) <- size.factors$SizeFactor

reducedDim(thymus.sce, "PCA") <- as.matrix(pca.dims[, paste0("PC", 1:50)])
thymus.sce

We are now in the position to save all of these data. We'll do this by breaking up the SingleCellExperiment object into smaller, sorting day-wise objects. as this will help to increase the speed of downloading. These smaller files are what get uploaded to r Biocpkg("ExperimentHub").

base <- file.path("MouseThymusAgeing", "SMARTseq", "1.0.0")
dir.create(base, recursive=TRUE, showWarnings=FALSE)
saveRDS(rowData(thymus.sce), file=paste0(base, "/rowdata.rds"))
for(day in unique(thymus.sce$SortDay)){
    sub <- thymus.sce[, thymus.sce$SortDay == day]
    saveRDS(counts(sub), 
        file=paste0(base, "/counts-processed-day", day, ".rds"))
    saveRDS(colData(sub), 
        file=paste0(base, "/coldata-day", day, ".rds"))
    saveRDS(sizeFactors(sub), 
        file=paste0(base, "/sizefac-day", day, ".rds"))
    saveRDS(reducedDims(sub), 
        file=paste0(base, "/reduced-dims-day", day, ".rds"))
}

Session information

sessionInfo()

MarioniLab/MouseThymusAgeing documentation built on Feb. 19, 2023, 11:24 a.m.