MiRAnorm (version 1.0.0)

miranorm: Adaptive algorithm to identify normalization genes.

Description

miranorm returns a list of suggested normalizaiton genes based on supplied data. Various output figures are also available if requested in the parameter list.

Usage

miranorm(data = dat, group = dat$Trt, max = 15, min = 3, method = "complex", dis.method = "Euclidean", hclust.method = "single", ct = 25, missing = 0, clustplot = TRUE, selected = 4, ggplot = TRUE, heatmap = TRUE, known.positives = NULL, suggested.list = NULL, exclude = NULL)

Arguments

data

Dataframe containing at a minimum: Sample, Gene, Ct, and Trt

group

Treatment allocation. This should be the same length as the number of rows in data.

max

When method is chosen as "complex", this determines the maximum selected size at which the stability evaluation is done.

min

When method is chosen as "complex", this determines the minimum selected size at which the stability evaluation is done.

method

Choice of "simple" or "complex". Simple runs a single pass of miranorm for suggested normalizing genes. Complex runs bootstrap samples across a range of sizes to compute a stability metric.

dis.method

Distance metric used to calculate pairwise distance between individual miRNAs across all samples. Currently "Euclidean" and "1-Cor" are implemented.

hclust.method

The agglomeration method used to group genes. Methods are the same as defined in the hclust function in the stats package and include "single", "average", "complete", and "ward.D2". "single" is recommended as it is more robust to small perturbations and tends to form the "chaining" phenomenon useful for defining normalizing genes.

ct

Cycle threshold values at or above this level are treated as NA for the purposes of determining normalization genes. Recommend the value be set to 25.

missing

Defines maximum percentage of samples missing for a given gene before that gene is excluded from dataset during normalization.

clustplot

"True" or "False" to output or suppress stability plot. Only applicable if method = "complex".

selected

How many adaptive normalizing genes to search for in panel. Note, actual number of genes found may be larger based on tree cut.

ggplot

"True" or "False" to output general raw data plots.

heatmap

"True or "False" to output or suppress heatmap plot.

known.positives

Names of miRNA that are known positive. These will be added automatically to the heatmap plot.

suggested.list

Names of miRNA that are user suggested normalizing miRNA. These will be added automatically to the heatmap plot.

exclude

List of miRNA to exclude from the selection process for HK genes, eg: known.positives should be included here.

Value

A list including the following: nmean, nmed, nlcv, lcv.gene, ncls, ncls.gene.nmean is the dataset normalized to the global mean.nmed is the dataset normalized to the global median.nlcv is the dataset normalized to the average of the 3 genes with the lowest CV.lcv.gene is the names of the 3 genes with used to normalize nlcvncls is the dataset normalized to the adaptive normalizing genes chosen from miranorm.ncls.gene is the names of the genes chosen as adaptive normalizing genes chosen from miranorm.

Examples

dat = simData(n.trt=15, n.ctrl=15, n.gene=30, n.err=10, sigma.error = c(1, 0.3), mean.sample = 2,
sigma.sample = 1.88 , sigma.gene = 0.1, n.big.effect = 5, n.small.effect = 10, mean.big.effect = 2,
mean.small.effect = 1.2)$sim

obj = miranorm(data = dat, group = dat$Group, method="simple")