diag_plots: QC/diagnostic plotting

Description Usage Arguments Value Examples

View source: R/plot_wrappers.R

Description

Wrappers for a series of plots to be used as diagnostics in RNA-seq analyses. Currently 10 plots are possible using this function: 1) Mapped reads, 2) Relative Log Expression (RLE), 3) Principle Component Analyis (PCA), 4) Residuals from a batch correction model, e.g. RUVseq, 5) Hierarchical clustering, 6) Densitiy distributions, 7) Boxplots, 8) MA plots, 9) Volcano Plots and 10) P-value distribution plots. Plots 1 to 6 utilise a "SeqExpressionSet" object for extracting information to plot. Plots 8-10 utilised a simple list class, containing all the data.frames of each comparison performed. See descriptions of each in the parameter options below and for format specification. See vignette for more information and examples.

Usage

1
2
3
4
5
diag_plots(se_in = NULL, merged_in = NULL, write = FALSE,
  plot_dir = NULL, legend = TRUE, label = TRUE, name = NULL,
  mapped_reads = FALSE, rle = FALSE, pca = FALSE,
  residuals = FALSE, hclust = FALSE, density = FALSE,
  boxplot = FALSE, ma = FALSE, volcano = FALSE, p_dist = FALSE)

Arguments

se_in

A "SeqExpressionSet" object or "RangedSummarizedExperiment" generated using "buildSummarized()". If the input is a "SeqExpressionSet", ensure that it included groups to be analysed. E.g. accessible as "se_in$group. Groupings are used to automate colouring of samples in unsupervised analyses. Default = NULL

merged_in

A data.frame that contains the merged results which are included in the outputs from multi_de_pairs(). These contain the ouputs from the pair-wise comparisons which allows plotting of MA, Volcano and p-value distributions. Where the outputs of multi_de_pairs() are to be used as inputs into diag_plots(), use multi_de_pairs()$merged as inputs. See example below. Default = NULL

write

Write the results to a pdf file? Options: TRUE, FALSE. This is to be used together with "plot_dir" and "write" parameters (below). Will report an error and halt if is TRUE and "plot_dir" and "write" are NULL. Default = FALSE

plot_dir

If "write" is TRUE, where to write the files to? The directory must already exist. E.g. "/path/to/my/pretty/plots/". Default = NULL

legend

Include legend in plots? Legend is based on group data in se_in. Options: TRUE, FALSE. Default = FALSE

label

Include point labels in plots? Points are based on ID column from merged_in. Options: TRUE, FALSE. Default = FALSE

name

If "write" is TRUE, what to name the plot? The file name will always be preceded with "QC_" and end in ".pdf". E.g. name="very_pretty_plots" will produce a file named "QC_very_pretty_plots.pdf" in "/path/to/my/pretty/plots/". Default = NULL

mapped_reads

Plot mapped reads per sample as a barchart. Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

rle

Plot Relative Log Expressio (RLE) of samples for assessment of sample quality. See ?plotRLE for further details. Requires se_in to be a "SeqExpressionSet"and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

pca

Perform unsupervised Principle Component Analysis (PCA) and plot results. By default performs Singular Value Decomposition. Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

residuals

If RUV-seq has been applied to dataset, plot the residuals identified in the model. Only works for one set of residuals. Data is also accessible using pData(se_in)$W_1. Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

hclust

Performs unsupervised hierarchical clustering of samples. Colours sample below plot according to group and numbered by inputs. Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

density

Plot density distributions of log2(count-per-million). Will automatically extract normalised counts over non-normalised counts is available in "SeqExpressionSet". Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

boxplot

Boxplot of density distributions of log2(count-per-million). Will automatically extract normalised counts over non-normalised counts is available in "SeqExpressionSet". Requires se_in to be a "SeqExpressionSet" and utilise "group" meta-data for colouring. Options: TRUE, FALSE. Default = FALSE

ma

Plot Mean versus. Log2 Fold-Change of comparison. Requires a data.frame as input to "merged_in" with the following column names "ID", "AvExpr", "Log2FC" and "Adj_PVal".The data frame should be sorted, as the top 10 in the table are also plotted. Options: TRUE, FALSE. Default = FALSE

volcano

Volcano plot of Log2 Fold-Change and significance of comparison. Requires a data.frame as input to "merged_in" with the following column names "ID", "AvExpr", "Log2FC" and "Adj_PVal". The data frame should be sorted, as the top 10 in the table are also plotted. Options: TRUE, FALSE. Default = FALSE

p_dist

P-value distribution plot. Requires a data.frame as input to "merged_in" with the following column names "ID", "AvExpr", "Log2FC" and "Adj_PVal". The data frame should be sorted, as the top 10 in the table are also plotted. Options: TRUE, FALSE. Default = FALSE

Value

Returns pretty plots

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
## Load the example data set and attach
## The example below will display a PCA plot before normalisation
library(airway)
data(airway)
## Name the groups of the data.
colData(airway)$group <- colData(airway)$dex
## Identify the file locations
colData(airway)$file <- rownames(colData(airway))
## Filter low count data:
airway_filter <- buildSummarized(summarized = airway,
                                 filter = TRUE)
## for illustration, use random sample of 1000 transcripts
set.seed(1234)
airway_filter <- sample(airway_filter, 1000)
## The following is example code to perform a PCA plot
## see vignette for more details of displaying each plot
## diag_plots(se_in = airway_filter,
##            name = "airway example data",
##            pca = TRUE)
              

awaardenberg/consensusDE documentation built on Dec. 4, 2019, 9:08 p.m.