qualityplots: Make different plots for microarray quality assessment.

In machalen/QualityGraphs: Make quality graphs for microarray data based on objects obtained witharoma affymetrix normalization.

Description

This function can be used as a shortcut to make all quality plots included in QualityGraphs package with a single call.

Usage

qualityplots(ds, picname, estimates_m = NULL, est_noctrls, labels = NULL, ds.N = NULL, plmTr = NULL, conditions = NULL, colors = NULL, groupn = NULL, all = FALSE, Density = TRUE, Boxplots = TRUE, Clusters = TRUE, PCA = TRUE, estimates = FALSE, noctrls = TRUE, resDir = NULL)

Arguments

ds	Object obtained with aroma.affymetrix and the function AffymetrixCelSet in the aroma.affymetrix package.
picname	Name of output files.
estimates_m	Numeric matrix where columns are samples and rows genes. estimates_m corresponds to the expression matrix with controls.
est_noctrls	Numeric matrix where columns are samples and rows genes. est_noctrls corresponds to the expression matrix without controls.
labels	Optional. Vector with sample names. Samples are in the same order as in the ds object. Default get sample names from the ds object. It only works for boxplots and density plots
ds.N	Aroma affymetrix ds.N object with the normalized data from .CEL files using RMA and quantile normalization.
plmTr	plmTr object obtained with the ExonRmaPlm function in aroma.affymetrix() package.
conditions	Optional, vector with sample condition in the same order as sample columns in the expression matrix. IMPORTANT sample columns and sample conditions have to be in order mixedsort(labels).
colors	Optional, vector with the colous corresponding to each condition in the same order as sample columns in the expression matrix. IMPORTANT sample columns and sample conditions have to be in order mixedsort(labels).
groupn	Optional. Number of samples to include in each density plot. If the number is lower than total number of samples, more than one plot is going to be shown. Default is groupn = all samples.
all	Optional. If TRUE, an additional plot with all density plots made with a groupn is made.
Density	Call densityplot_all function to make density plots. Default is TRUE.
Boxplots	Call boxplot_all function to make boxplots. Default is TRUE.
Clusters	Call clusterdend function to make cluster dendrograms. Default is TRUE.
PCA	Call makePCA function to make PCA plots. Default is TRUE.
estimates	If TRUE, clusters for estimates_m are made. Default is FALSE.
noctrls	If TRUE, clusters for est_noctrls are made. Default is TRUE.
resDir	Output results directory. Default is ResultsDir.

Details

This function is a shortcut to call functions: makePCA, clusterdend, boxplot_all and densityplot_all. It makes all the plots for quality assessment in microarray data. Density plots show the distribution of log2 raw intensities to check possible technical biases. Boxplots show the log2 raw intensities in boxplots to check possible technical biases. BoxplotsRMA show normalized log2 intensities with quantile normalization and Robust Multiarray average (RMA). It is used to check if normalization is made properly. Boxplot NUSE plot with normalized unscaled standard errors (NUSE). The standard error from the probe-level model are visualized as boxplots. The deviant arrays can be identified by not being centered at 1, or being more spread out than the other arrays.Boxplot RLE plot with relative log expression values (RLE). Assuming that most probesets on arrays are not changing, most of the RLE values are close 0. When examining the boxplot, the deviant arrays can be identified by not being centered at 0, or being more spread out than the other arrays.Clusters show the sample distribution using different methods. Samples are expected to group together based on their similarity. PCA show the sample distribution using a principal component analysis (PCA) in a 3D plot.

Value

Files are created in the resDir directory with the picname and .png or .pdf extensions.

Author(s)

Magdalena Arnal Segura <[e-mail]>.

machalen/QualityGraphs documentation built on Oct. 22, 2019, 8:29 p.m.