qcUniformClusters: Computing of the fraction of cluster with uniform phenotypes

In tchitchek-lab/SPADEVizR: SPADEVizR: an R package for Visualization, Analysis and Integration of SPADE results

Computing of the fraction of cluster with uniform phenotypes

Description

The 'qcUniformClusters()' function allows to report as PDF files 2 kinds of information. Firstly, the identification of clusters having non-unimodal marker expression densities is performs using a Hartigan's dip test. Secondly, markers of cluster having a large spread of expressions (above a threshold) are identified.

If all the clustering marker expressions of a cluster checks this assessments (depending of the 'uniform.test' parameter), the cluster is considered as uniform.

Usage

qcUniformClusters(Results, clusters = NULL,
  only.clustering_markers = FALSE, uniform.test = "both",
  th.pvalue = 0.05, th.IQR = 2,
  density.PDFfile = "qcUniformClusters_density.pdf",
  density.PDFfile.dim = c(17, 10),
  heatmap.PDFfile = "qcUniformClusters_heatmap.pdf",
  heatmap.PDFfile.dim = c(length([e-mail]),
  length([e-mail])/4), tile.color = "black",
  verbose = TRUE, ...)

Arguments

Results	a Results object (with 'flowset' slot not null)
clusters	a character vector containing the clusters names to be computed (by default all clusters will be computed)
only.clustering_markers	a logical specifying if only clustering marker densities must be displayed.
uniform.test	a character specifying the qc assessment to perform
th.pvalue	a numeric value specifying the pvalue threshold of the Hartigan's dip test (multimodal if p.value < th.pvalue)
th.IQR	a numeric value specifying the IQR (interquartile range) threshold to assume a distribution as uniform
density.PDFfile	a character specifying the output path of the marker expression density plots (or NULL to avoid this step)
density.PDFfile.dim	a numeric vector specifying the width and the height of the density PDF file
heatmap.PDFfile	a character specifying the output path of the marker expression accuracy heatmap (or NULL to avoid this step)
heatmap.PDFfile.dim	a numeric vector specifying the width and the height of the heatmap PDF file
tile.color	a character specifying the border color of the tiles (NA to remove tile borders)
verbose	a boolean specifying if some messages must be displayed during the generation of the qc report
...	supplemental parameters passed to the dip.test function

Details

The test of multimodal distribution is performs using a Hartigan's dip test. The null hypothesis of this test assume the distribution as unimodal Supplemental parameters passed to the dip.test function are ('simulate.p.value' and 'B'). 'simulate.p.value' is a logical indicating whether to compute p-values by Monte Carlo simulation (FALSE by default). 'B' is an integer specifying the number of replicates used in the Monte Carlo test (2000 by default).

The spread of a marker expression is considered as high if the IQR (interquartile range) is higher than a provided threshold th.IQR.

The 'uniform.test' parameter can take 3 values: "unimodality": check if the distribution of the marker expression is unimodal or not using a Hartigans dip test "spread": check if the spread of the marker expression is below an IQR threshold (interquantile range) "both": check "unimodality" and "spread"

Value

a list containing a numeric (perc numeric element) specifying the percentage of clusters having only uniform clustering marker expressions and a dataframe (accuracy.matrix element) specifying if each marker of each cluster is uniform or not.

tchitchek-lab/SPADEVizR documentation built on Jan. 27, 2024, 8:58 p.m.