In LTLA/scRNAseq: Collection of Public Single-Cell RNA-Seq Datasets

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

Promoting to `SingleCellExperiment`

For some reason, I saved the alternative experiment as a SummarizedExperiment instead of a SingleCellExperiment. I was probably hoping to eliminate the potential for unnecessary recursion of alternative experiments, but in doing so, got rid of useful features for SingleCellExperiments like reduced dimensions and size factors.

So let's put that stuff back. First we clone the project:

tmp <- "update_20240418"
unlink(tmp, recursive=TRUE) # get rid of any previous directory.
gypsum::cloneVersion("scRNAseq", "mair-pbmc-2020", "2023-12-20", destination=tmp)

Now we go in and promote the OBJECT file to a SingleCellExperiment, following the specification here. There's no need to do anything else as the new SCE just differs from the old SE by the class definition, not any of its contents.

update <- function(obj.path) {
    obj <- jsonlite::fromJSON(obj.path, simplifyVector=FALSE)
    obj$type <- "single_cell_experiment"
    obj$ranged_summarized_experiment <- list(version="1.0")
    obj$single_cell_experiment <- list(version="1.0")

    # Replacing the link with the new content; it's important to delete the link
    # before doing so, otherwise we would accidentally modify the cache!
    unlink(obj.path) 
    write(jsonlite::toJSON(obj, auto_unbox=TRUE, pretty=4), file=obj.path)
}

Running our update script on all of the alternative experiments:

update(file.path(tmp, "alternative_experiments/0", "OBJECT"))

We validate and load the entire object, just to check that it's all good.

library(alabaster.base)
validateObject(file.path(tmp))
out <- readObject(tmp)
ae <- SingleCellExperiment::altExp(out)
stopifnot(is(ae, "SingleCellExperiment"))
ae