For the examples in this book, we will rely on a set of publicly available datasets that cover different sequencing-based and imaging-based platforms, namely: Visium, Visium HD, Xenium (10x Genomics), and CosMx (Bruker).
This chapter provides an overview of the example datasets used in the code examples in the later chapters.
These datasets have been deposited in an Open Storage Framework (OSF) repository here, and can be easily queried and downloaded using functions from the osfr CRAN package. For convenience, we have implemented the OSTA.data Bioconductor package to:
It can happen that we change, add, or remove data from the OSF repositorying underlying OSTA.data. Should you ever run into any issues that might be related to this, we suggest removing affected cache resource(s) as follows, and retrieving these data anew.
library(BiocFileCache)
bfc <- BiocFileCache()
# specify dataset identifier
id <- "Xenium_HumanColon_Oliveira"
# query cached files for 'id'
que <- bfcquery(bfc, id)
# clear matching resource
bfcremove(bfc, que$rid)
# retrieve current dataset
OSTA.data_load(id)polygons and molecules to skip these dataThe following datasets are currently available:
## [1] "Chromium_HumanBreast_Janesick" "Chromium_HumanColon_Oliveira"
## [3] "CosMx1k_MouseBrain1" "CosMx1k_MouseBrain2"
## [5] "CosMx6k_HumanBrain" "VisiumHD_HumanColon_Oliveira"
## [7] "Visium_HumanBreast_Janesick" "Visium_HumanColon_Oliveira"
## [9] "Xenium_HumanBreast1_Janesick" "Xenium_HumanColon_Oliveira"In addition, several datasets are available from the STexampleData Bioconductor package as pre-formatted SpatialExperiment and SingleCellExperiment formats. These data objects are stored on Bioconductor’s ExperimentHub resource, and can be loaded in R by querying ExperimentHub or using loader functions provided in the STexampleData package.
library(STexampleData)
# show help file listing datasets
# ?STexampleDataData files downloaded with the packages above are stored and managed in a temporary directory using BiocFileCache. Sometimes, these temporary files may need to be deleted manually, for example if there has been a recent change to files stored in the OSF repository. The code example below shows how to remove the temporary files for one of these datasets. Alternatively, you can also find the temporary directory on your system with BiocFileCache::BiocFileCache() and delete files individually.
# locate and delete files in BiocFileCache directory
id <- "VisiumHD_HumanColon_Oliveira"
bfc <- BiocFileCache::BiocFileCache()
qid <- BiocFileCache::bfcquery(bfc, id)$rid
BiocFileCache::bfcremove(bfc, qid)Below, we briefly summarize the characteristics of several key datasets, and note across which parts of the book these are being used.
In the underlying paper (Janesick et al. 2023), the Xenium data (2 replicates) were accompanied by consecutive slices of Chromium and Visium data. Therefore, these replicates are expected to have nearly identical biological findings. By transferring Chromium cell type labels to spatial technologies, such as Visium (with full transcriptome) and Xenium (at single-cell resolution), we can combine analytical insights from different platforms.
In the underlying paper, there are both normal adjacent tissue (NAT) and colorectal carcinoma (CRC) samples from 5 patients. The Visium HD data (P2 CRC) were accompanied by consecutive slices of Chromium, Visium, and Xenium data. Therefore, we can jointly analyze these modalities.
There are two sections from the CosMx mouse brain sample, namely “coronal hemisphere” (sample 1) and “coronal hippocampus and cortex” (sample 2).
The CosMx human prefrontal cortex sample has a larger gene panel of ~6,000 RNA targets.