topHitsByStudy {MetaNeighbor}R Documentation

Find reciprocal top hits, stratifying results by study.

Description

This function looks for reciprocal top hits in each target study separately, allowing for as many reciprocal top hits as target studies. This is the recommended function for extracting top hits.

Usage

topHitsByStudy(
  auroc,
  threshold = 0.9,
  n_digits = 2,
  collapse_duplicates = TRUE
)

Arguments

auroc

matrix of celltype-to-celltype AUROC scores (output from MetaNeighborUS)

threshold

AUROC threshold, must be between [0,1]. Default is 0.9. Only top hits above this threshold are included in the result table.

n_digits

Number of digits for AUROC values in the result table. Set to "Inf" to skip rounding.

collapse_duplicates

Collapse identical pairs of cell types (by default), effectively averaging AUROCs when reference and target roles are reversed. Setting this option to FALSE makes it easier to filter results by study or cell type. If collapse_duplicates is set to FALSE, "Celltype_1" is the reference cell type and "Celltype_2" is the target cell type (relevant if MetaNeighborUS was run with symmetric_output = FALSE).

Value

Function returns a dataframe with cell types that are either reciprocal best matches, and/or those with AUROC values greater than or equal to threshold value

See Also

topHits

Examples

data(mn_data)
var_genes = variableGenes(dat = mn_data, exp_labels = mn_data$study_id)
aurocs = MetaNeighborUS(var_genes = var_genes, 
                        dat = mn_data, 
                        study_id = mn_data$study_id,
                        cell_type = mn_data$cell_type)
top_hits = topHitsByStudy(aurocs)
top_hits


[Package MetaNeighbor version 1.13.0 Index]