Build and plot dendrogram from gene panel — buildTreeFromGenePanel • mfishtools

Build and plot a dendrogram using correlation-based average linkage hierarchical clustering and only using a specified set of genes. The output is the expected accuracy of mapping to each node in the tree, which gives an idea of the best-case expected results for mFISH analysis.

buildTreeFromGenePanel(
  dend = NA,
  refDat = NA,
  mapDat = refDat,
  medianDat = NA,
  requiredGenes = 2,
  clusters = NA,
  mappedAsReference = FALSE,
  genesToMap = rownames(mapDat),
  plotdendro = TRUE,
  returndendro = TRUE,
  mar = c(12, 5, 5, 5),
  main = NULL,
  ylab = NULL,
  use = "p",
  ...
)

Arguments

dend: dendrogram for mapping. Ignored if medianDat is passed
refDat: normalized data of the REFERENCE data set. Ignored if medianDat is passed
mapDat: normalized data of the MAPPING data set. Default is to map the data onto itself.
medianDat: representative value for each leaf and node. If not entered, it is calculated
requiredGenes: minimum number of genes required to be expressed in a cluster (column of medianDat) for the cluster to be included (default=2)
clusters: cluster calls for each cell
mappedAsReference: if TRUE, returns the fraction of cells mapped to a node which are were orginally clustered from that node; if FALSE (default) returns the fraction of cells clustered under a node which are mapped to the correct node.
genesToMap: which genes to include in the correlation mapping
plotdendro: should the dendrogram be plotted (default = TRUE)
mar: margins (for use with par)
main, ylab: add title and labels to plot (default is NULL)
use, ...: additional parameters for cor
returnDendro: should the dendrogram be returned (default = TRUE)

Value

a list where the first entry is the resulting tree and the second entry is the fraction of cells correctly mapping to each node using the inputted gene panel.