hcaquery

• This package, in conjunction with the hca package, allows for exploration of the Human Cell Atlas' data, made available by their API.

• Its core functionality is the generation of relational database tables containing data from .loom and .h5ad files

• This is done by using the hca package to obtain .loom and .h5ad files from the HCA API, each of which can be parse into a SingleCellExperiment object, providing us with access to (1) cell annotations, (2) gene annotations, (3) assay matrices, and (4) file metadata.

• Each of these data categories in turn corresponds to a table in the relational database schema:

  1. genes_tbl
  2. cells_tbl
  3. assays_tbl
  4. experiment_overviews

Examples (in development)

devtools::load_all()
library(hca)
library(dplyr)

• loading the two smallest .loom files

loom_filter <- hca::filters(fileFormat = list(is = c("loom")))
loom_tbl <- hca::files(filters = loom_filter,
                            size = 2, sort = "fileSize", order = "asc")
files_to_db(loom_tbl)

• loading .loom files produced by a single project, and processed by the HCA

filters <- filters(
    fileFormat = list(is = "loom"),
    fileSource = list(is = "DCP/2 Analysis")
)
files <- files(filters) # how many? 46 files

files |>  # number and total size per project
    group_by(projectTitle) |>
    summarize(n = n(), GB = sum(size) / (1024^3)) |>
    arrange(GB)

filters <- filters(
    ## a particular project, with 3 files, 1/2 GB
    projectId = list(is = "88ec040b-8705-4f77-8f41-f81e57632f7d"),
    fileFormat = list(is = "loom"),
    fileSource = list(is = "DCP/2 Analysis")
)

loom_tbl <- files(filters)
files_to_db(loom_tbl)