get.consensus.subtypes: Get consensusOV ovarian cancer subtypes

In consensusOV: Gene expression-based subtype classification for high-grade serous ovarian cancer

Description

Get consensusOV ovarian cancer subtypes

Usage

get.consensus.subtypes(expression.matrix, entrez.ids,
  concordant.tumors.only = TRUE, remove.using.cutoff = FALSE,
  percentage.dataset.removed = 0.75,
  .training.dataset = consensus.training.dataset.full,
  .dataset.names.to.keep = names(esets.rescaled.classified.filteredgenes))

margin(rf.probs)

Arguments

expression.matrix	A matrix of gene expression values with rows as genes, columns as samples.
entrez.ids	A vector of Entrez Gene IDs, corresponding to the rows of expression.matrix
concordant.tumors.only	Logical. Should the classifier trained only on tumors that are concordantly classified by Helland, Konecny, and Verhaak? Defaults to TRUE.
remove.using.cutoff	Specify whether to classify NA for samples that do not meet a margin cutoff
percentage.dataset.removed	If remove.using.cutoff is TRUE, then classify this percentage of samples to NA based on margin values
.training.dataset	ExpressionSet containing the training data. Defaults to the pooled dataset across selected MetaGxOvarian datasets.
.dataset.names.to.keep	Names of MetaGxOvarian datasets to use for training
rf.probs	random forest probabilities for each subtype as returned by get.consensus.subtypes

Value

get.consensus.subtypes returns a list with first value consensusOV.subtypes containing a factor of subtype labels; and second value rf.probs containing a matrix of subtype probabilities.

margin returns a numeric vector containing the classification margin scores, i.e. the difference between the top two subtype scores for each tumor.

Examples

library(Biobase)
data(GSE14764.eset)
expression.matrix <- exprs(GSE14764.eset)
entrez.ids <- as.character(fData(GSE14764.eset)$EntrezGene.ID)
sts <- get.consensus.subtypes(expression.matrix, entrez.ids)
margins <- margin(sts$rf.probs)

consensusOV documentation built on Nov. 8, 2020, 7:06 p.m.