markerCandidates: Write candidate marker genes to a file

Description Usage Arguments

View source: R/markerGeneSelection.R

Description

Writes candidate marker genes to a file along with other required information (fold change and silhouette coefficient) Selects candidates based on fold change to the median expression of other samples and a minimum expression in the cell type

Usage

1
2
3
4
5
6
markerCandidates(design, expression, outLoc, groupNames,
  regionNames = NULL, PMID = "PMID", rotate = NULL, cores = 1,
  debug = NULL, sampleName = "sampleName",
  replicates = "originalIndex", foldChangeThresh = 10,
  minimumExpression = 8, background = 6, regionHierarchy = NULL,
  geneID = "Gene.Symbol", seed = NULL)

Arguments

design

data.frame. Metadata for the samples.

expression

data.frame. Expression data for the samples. Gene names should be included as a column. Any other non expression data not be of type double

outLoc

Directory to save the candidate genes

groupNames

column of the design with cell type names. If multiple columns are provided, selection will be performed for each independently

regionNames

column of the design with region names. Multiple regions can be listed separated by commas

PMID

column of the design with pubmed identifiers. This is required to identify the samples coming from the same study

rotate

double. percentage of samples to be removed. 0.33 is used for the study

cores

Number of cores to use for paralellization

debug

Nothing to see here

sampleName

column of the design with sample names matching the column names of the expression data

replicates

column of the design that groups replicates

foldChangeThresh

minimum fold change required for selection

minimumExpression

minimum level of expression for a marker gene in its cell type

background

level of expression that should be considered as background level

regionHierarchy

hiearchy of regions.

seed

seed for random generation. if NULL will be set to random


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