Aggregate cell types into groups — aggregate_cell

Adds, for every method-signature combination, a new feature <method>_<signature>_<group> with the sum of the cell types of the group (e.g. Myeloid cells = macrophages + monocytes + dendritic cells). Cell types are only summed within the same method-signature combination, and the original features are kept.

Usage

aggregate_cell_groups(
  deconvolution,
  cell_groups,
  min_types = 2,
  verbose = TRUE
)

Arguments

deconvolution: Deconvolution output of compute.deconvolution() with features as columns and samples as rows
cell_groups: A named list: each name is a group and each element a character vector with the cell types to sum, written as in get_cell_type_nomenclature() (e.g. list(Myeloid.cells = c("Macrophages.M1", "Monocytes"))). A group name can be a cell type of the nomenclature (e.g. Myeloid.cells): combinations that already estimate it are left as they are. Other group names must not contain _ nor the name of a cell type of the nomenclature. The cell types of a group should not overlap (do not list a cell type together with its own subtypes).
min_types: Minimum number of cell types of the group that a method-signature combination must have to be aggregated. Combinations with fewer are skipped (with 1, the group would be a copy of a single cell type).
verbose: Boolean value to whether print the cell types summed in each method-signature combination and the reminder of how to use the groups in the other functions

Value

The deconvolution matrix with the group features added as new columns.

Details

It is an optional step after compute.deconvolution(): the returned matrix can be used in the other functions as any other deconvolution matrix. Two things are needed for the groups to be used there (the function prints them as a reminder):

cells_extra: group names that are not in the nomenclature (e.g. Lymphocytes) must be given in cells_extra to compute.deconvolution.analysis(), compute.benchmark() and prepare_multideconv_folds(), otherwise these groups are discarded. Listing all the group names in cells_extra is always safe: names that are already in the nomenclature (e.g. Myeloid.cells) are ignored there.
New cohorts: before replicate_deconvolution_subgroups(), aggregate the same groups in the new deconvolution matrix. Otherwise the group features are missing in it and are set to NA.

Examples


data("deconvolution")

groups = list(Myeloid.cells = c("Macrophages.cells", "Macrophages.M0", "Macrophages.M1", "Macrophages.M2",
                                "Monocytes", "Dendritic.cells"),
              Lymphocytes = c("B.cells", "CD4.cells", "CD8.cells", "NK.cells"))

deconvolution_groups = aggregate_cell_groups(deconvolution, cell_groups = groups)
#> 
#> Group 'Myeloid.cells'
#>   Quantiseq: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   DeconRNASeq_BPRNACan: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   Epidish_BPRNACan: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   CBSX_BPRNACan: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DWLS_BPRNACan: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DeconRNASeq_BPRNACan3DProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   Epidish_BPRNACan3DProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   CBSX_BPRNACan3DProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DWLS_BPRNACan3DProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DeconRNASeq_BPRNACanProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   Epidish_BPRNACanProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   CBSX_BPRNACanProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DWLS_BPRNACanProMet: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DeconRNASeq_BSeqSC.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   Epidish_BSeqSC.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   CBSX_BSeqSC.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DWLS_BSeqSC.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DeconRNASeq_CBSX.HNSCC.scRNAseq: Macrophages.cells + Dendritic.cells
#>   Epidish_CBSX.HNSCC.scRNAseq: Macrophages.cells + Dendritic.cells
#>   CBSX_CBSX.HNSCC.scRNAseq: Macrophages.cells + Dendritic.cells
#>   DWLS_CBSX.HNSCC.scRNAseq: Macrophages.cells + Dendritic.cells
#>   DeconRNASeq_CBSX.Melanoma.scRNAseq: skipped (1 member: Macrophages.cells)
#>   Epidish_CBSX.Melanoma.scRNAseq: skipped (1 member: Macrophages.cells)
#>   CBSX_CBSX.Melanoma.scRNAseq: skipped (1 member: Macrophages.cells)
#>   DWLS_CBSX.Melanoma.scRNAseq: skipped (1 member: Macrophages.cells)
#>   DeconRNASeq_CBSX.NSCLC.PBMCs.scRNAseq: skipped (1 member: Monocytes)
#>   Epidish_CBSX.NSCLC.PBMCs.scRNAseq: skipped (1 member: Monocytes)
#>   CBSX_CBSX.NSCLC.PBMCs.scRNAseq: skipped (1 member: Monocytes)
#>   DWLS_CBSX.NSCLC.PBMCs.scRNAseq: skipped (1 member: Monocytes)
#>   DeconRNASeq_CBSX.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   Epidish_CBSX.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   CBSX_CBSX.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DWLS_CBSX.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DeconRNASeq_CCLE.TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   Epidish_CCLE.TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   CBSX_CCLE.TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   DWLS_CCLE.TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   DeconRNASeq_DWLS.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   Epidish_DWLS.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   CBSX_DWLS.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DWLS_DWLS.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DeconRNASeq_MOMF.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   Epidish_MOMF.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   CBSX_MOMF.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DWLS_MOMF.Vanderbilt.scRNAseq: already has Myeloid.cells (kept as it is)
#>   DeconRNASeq_TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   Epidish_TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   CBSX_TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   DWLS_TIL10: Macrophages.M1 + Macrophages.M2 + Monocytes + Dendritic.cells
#>   AutogeneS_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   BayesPrism_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   Bisque_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   CPM_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   MuSic_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   SCDC_Vanderbilt: already has Myeloid.cells (kept as it is)
#>   DeconRNASeq_LM22: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   Epidish_LM22: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   CBSX_LM22: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#>   DWLS_LM22: Macrophages.M0 + Macrophages.M1 + Macrophages.M2 + Monocytes
#> 
#> Group 'Lymphocytes'
#>   Quantiseq: B.cells + CD8.cells + NK.cells
#>   DeconRNASeq_BPRNACan: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_BPRNACan: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_BPRNACan: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_BPRNACan: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_BPRNACan3DProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_BPRNACan3DProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_BPRNACan3DProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_BPRNACan3DProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_BPRNACanProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_BPRNACanProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_BPRNACanProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_BPRNACanProMet: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_BSeqSC.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_BSeqSC.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_BSeqSC.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_BSeqSC.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_CBSX.HNSCC.scRNAseq: B.cells + CD4.cells + CD8.cells
#>   Epidish_CBSX.HNSCC.scRNAseq: B.cells + CD4.cells + CD8.cells
#>   CBSX_CBSX.HNSCC.scRNAseq: B.cells + CD4.cells + CD8.cells
#>   DWLS_CBSX.HNSCC.scRNAseq: B.cells + CD4.cells + CD8.cells
#>   DeconRNASeq_CBSX.Melanoma.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_CBSX.Melanoma.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_CBSX.Melanoma.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_CBSX.Melanoma.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_CBSX.NSCLC.PBMCs.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_CBSX.NSCLC.PBMCs.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_CBSX.NSCLC.PBMCs.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_CBSX.NSCLC.PBMCs.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_CBSX.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_CBSX.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_CBSX.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_CBSX.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_CCLE.TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_CCLE.TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_CCLE.TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_CCLE.TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_DWLS.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_DWLS.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_DWLS.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_DWLS.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_MOMF.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_MOMF.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_MOMF.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_MOMF.Vanderbilt.scRNAseq: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Epidish_TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CBSX_TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DWLS_TIL10: B.cells + CD4.cells + CD8.cells + NK.cells
#>   AutogeneS_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   BayesPrism_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   Bisque_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   CPM_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   MuSic_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   SCDC_Vanderbilt: B.cells + CD4.cells + CD8.cells + NK.cells
#>   DeconRNASeq_LM22: skipped (1 member: CD8.cells)
#>   Epidish_LM22: skipped (1 member: CD8.cells)
#>   CBSX_LM22: skipped (1 member: CD8.cells)
#>   DWLS_LM22: skipped (1 member: CD8.cells)
#> 
#> To use the aggregated matrix in other functions:
#>   - compute.deconvolution.analysis(), compute.benchmark(), prepare_multideconv_folds(): add cells_extra = "Lymphocytes", otherwise these groups are discarded
#>   - replicate_deconvolution_subgroups(): aggregate the same groups in the new cohort first, otherwise their features are set to NA