Skip to contents

Get SE colData and rowData

Source: R/se_to_rowcoldata.R
se_to_rowcoldata.Rd

Get SE colData and rowData consistently for SummarizedExperiment, SingleCellExperiment, Seurat data, and generic Biobase eSet compatible objects that provide featureData() and phenoData().

Usage

se_to_rowcoldata(se, verbose = FALSE, ...)

Arguments

se

recognized object:

  • SummarizedExperiment or any object that inherits from this class

  • SingleCellExperiment because it inherits from SummarizedExperiment

  • Seurat which is converted to SingleCellExperiment

  • ExpressionSet and any object that inherits from this class, using Biobase::featureData() and Biobase::phenoData().

  • NanoStringGeoMxSet which uses GeomxTools::sData() or NanoStringNCTools::sData() to define colData_se, otherwise is handled equivalent to ExpressionSet.

...

additional arguments are ignored

Value

list with two components:

  • "colData_se" as a data.frame with column metadata from se

  • "rowData_se" as a data.frame with row metadata from se

Details

This function provides a straightforward way to return the equivalent of data.frame(check.names=FALSE, rowData(se)) and data.frame(check.names=FALSE, colData(se)) for several types of object types.

  • It also defines rownames() and colnames() if either are missing.

  • When rowData has no annotation columns, it defines one column "rows" using rownames(se).

  • If slot name "rowRanges" exists, and "rowData" either does not exist or has zero columns, it will use rowRanges().

  • When colData has no annotation columns, it defines one column "columns" using colnames(se).

  • When input se class is "Seurat", it converts the object with Seurat::as.SingleCellExperiment()

  • Any other object uses Biobase::featureData() for "rowData_se", and Biobase::phenoData() for "colData_se".

To verify the logic used at each step, set verbose=TRUE.

For Class "NanoStringGeoMxSet" defined in GeomxTools, the colData_se is defined using GeomxTools::sData() in order to return the combined data.frame with protocolData() and pData() together. Accordingly, it is possible to have duplicated colnames, which becomes a problem for many downstream tools, so the second instance of any duplicated colnames have "_v1" added by using jamba::makeNames(x, renameFirst=FALSE). The first instance of each duplicated colname is not renamed. Use verbose=TRUE to confirm when duplicated columns are detected and renamed.

See also

Other jamses SE utilities: geomx_to_se(), make_se_test(), se_collapse_by_column(), se_collapse_by_row(), se_detected_rows(), se_normalize(), se_rbind(), se_to_assay_data(), se_to_assay_names(), simulate_se_test()