plotDiffHeatmap: Plot differential heatmap

In HelenaLC/CATALYST: Cytometry dATa anALYSis Tools

Plot differential heatmap

Description

Heatmaps summarizing differental abundance & differential state testing results.

Usage

plotDiffHeatmap(
  x,
  y,
  k = NULL,
  top_n = 20,
  fdr = 0.05,
  lfc = 1,
  all = FALSE,
  sort_by = c("padj", "lfc", "none"),
  y_cols = list(padj = "p_adj", lfc = "logFC", target = "marker_id"),
  assay = "exprs",
  fun = c("median", "mean", "sum"),
  normalize = TRUE,
  col_anno = TRUE,
  row_anno = TRUE,
  hm_pal = NULL,
  fdr_pal = c("lightgrey", "lightgreen"),
  lfc_pal = c("blue3", "white", "red3")
)

Arguments

x	a SingleCellExperiment.
y	a SummarizedExperiment containing differential testing results as returned by one of testDA_edgeR, testDA_voom, testDA_GLMM, testDS_limma, or testDS_LMM. Alternatively, a list as returned by diffcyt.
k	character string specifying the clustering in x from which y was obtained. If NULL, plotDiffHeatmap will try and guess it, which will be inaccurate if multiple clusterings share the same levels.
top_n	numeric. Number of top clusters (if type = "DA") or cluster-marker combinations (if type = "DS") to display.
fdr	numeric threshold on adjusted p-values below which results should be retained and considered to be significant.
lfc	numeric threshold on logFCs above which to retain results.
all	logical specifying whether all top_n results should be displayed. If TRUE, fdr,lfc filtering is skipped.
sort_by	character string specifying the y column to sort by; "none" to retain original ordering. Adj. p-values will increase, logFCs will decreasing from top to bottom.
y_cols	named list specifying columns in y that contain adjusted p-values (padj), logFCs (lfc) and, for DS results, feature names (target). When only some y_cols differ from the defaults, specifying only these is sufficient.
assay	character string specifying which assay data to use; valid values are assayNames(x).
fun	character string specifying the function to use as summary statistic for aggregation of assay data.
normalize	logical specifying whether Z-score normalized values should be plotted. If y contains DA analysis results, frequencies will be arcsine-square-root scaled prior to normalization.
col_anno	logical specifying whether to include column annotations for all non-numeric cell metadata variables; or a character vector in names(colData(x)) to include only a subset of annotations. (Only variables that map uniquely to each sample will be included)
row_anno	logical specifying whether to include a row annotation indicating whether cluster (DA) or cluster-marker combinations (DS) are significant, labeled with adjusted p-values, as well as logFCs.
hm_pal	character vector of colors to interpolate for the heatmap. Defaults to brewer.pal's "RdYlBu" for DS, "RdBu" for DA results heatmaps.
fdr_pal, lfc_pal	character vector of colors to use for row annotations fdr_pallength 2 for (non-)significant at given fdr lfc_pallength 3 for negative, zero and positive

Value

a Heatmap-class object.

Author(s)

Lukas M Weber & Helena L Crowell [e-mail]

Examples

# construct SCE & run clustering
data(PBMC_fs, PBMC_panel, PBMC_md)
sce <- prepData(PBMC_fs, PBMC_panel, PBMC_md)
sce <- cluster(sce, verbose = FALSE)

## differential analysis
library(diffcyt)

# create design & constrast matrix
design <- createDesignMatrix(ei(sce), cols_design=2:3)
contrast <- createContrast(c(0, 1, 0, 0, 0))

# test for
# - differential abundance (DA) of clusters
# - differential states (DS) within clusters

da <- diffcyt(sce, design = design, contrast = contrast, 
    analysis_type = "DA", method_DA = "diffcyt-DA-edgeR",
    clustering_to_use = "meta20", verbose = FALSE)
    
ds <- diffcyt(sce, design = design, contrast = contrast, 
    analysis_type = "DS", method_DS = "diffcyt-DS-limma",
    clustering_to_use = "meta20", verbose = FALSE)
    
# extract result tables
da <- rowData(da$res)
ds <- rowData(ds$res)
    
# display test results for
# - top DA clusters
# - top DS cluster-marker combinations
plotDiffHeatmap(sce, da)
plotDiffHeatmap(sce, ds)

# visualize results for subset of clusters
sub <- filterSCE(sce, cluster_id %in% seq_len(5), k = "meta20")
plotDiffHeatmap(sub, da, all = TRUE, sort_by = "none")

# visualize results for selected feature
# & include only selected annotation
plotDiffHeatmap(sce["pp38", ], ds, col_anno = "condition", all = TRUE)

HelenaLC/CATALYST documentation built on Nov. 30, 2024, 4:04 a.m.