mgpEstimate: Calculate cell type profile estimations

In PavlidisLab/markerGeneProfile: Cell type marker gene profiles in whole tissue samples

Description

Primary function for cell type profile estimations.

Usage

mgpEstimate(exprData, genes, geneColName = "Gene.Symbol",
  outlierSampleRemove = FALSE, geneTransform = function(x) {    
  homologene::mouse2human(x)$humanGene }, groups = NULL,
  tableOut = NULL, indivGenePlot = NULL, seekConsensus = FALSE,
  removeMinority = TRUE, plotType = c("groupBased", "cummulative"),
  permuations = 0, PC = 1)

Arguments

exprData	data.frame. Expression data. First collumns of the expression data should include gene names in the same format as the ones specified in the marker gene lists. Any other non-expression related fields must not be of type 'double'
genes	a named list containing marker gene lists of each cell type
geneColName	character. name of the column containing the gene names in the expression file
outlierSampleRemove	logical. Deprecated. If TRUE, outlier samples will be removed from the output. Outliers are calculated in the context of a group
geneTransform	a function that will be applied to the gene list. the default behavior is to change mouse genes to human genes. set to NULL to keep the genes as they are
groups	a vector stating which groups each sample belongs to
tableOut	character, directory name. If provided outputs loadings of individual genes and variance explained by principal components
indivGenePlot	character, directory name. If provided, plots expression of marker genes in individual groups per marker gene list. Is not guaranteed to look pretty.
seekConsensus	logical. If TRUE any gene with negative loadings in any of the groups individually will be removed. Use if there is a high likelihood of gene regulation between the groups.
removeMinority	logical. should the genes with negative loadings be removed from the estimation. Setting seekConsensus to TRUE makes this irrelevant. As all negatives will be removed at that step
plotType	if indivGenePlot is provided, type of plot to be saved. groupBased separates expression between groups cummulative plots a single value
PC	which principal component to use. For debugging purposes. Recommended value is always 1

Value

A list.

• estimates. A named vector of marker gene profiles estimated bythe function.

• groups. if provided this lists the groups that the samples belong to. Unless outlierSampleRemove is set to TRUE, this will be the same as the groups for all cell types. So in most cases ignore this.

• rotations. Loadings from the PCA. These values are modified to ensure the chosen PC (1st by default) has a positive direction.

• trimmedPCAs. Raw PCA object used to calculate the estimates. Rotation direction is not changed

• fullPCAs. Raw PCA object with all marker genes before their removal by removeMinority and seekConsensus options.

• removedMarkerRatios. Ratio of markers removed by removeMinority and seekConsensus options.

PavlidisLab/markerGeneProfile documentation built on July 11, 2019, 11:42 p.m.