multi_de_pairs: Batch - multiDE analysis of many comparisons

Description Usage Arguments Value Examples

View source: R/multi_DE.R

Description

Given a summarized experiment generated using buildSummarized() this function will automatically perform differential expression (DE) analysis for all possible groups using 3 different methods 1) EdgeR, 2) Voom and 3) DEseq2. It will also output 10x diagnostic plots automatically, if the plotting options are selected (see ?diag_plots for more details).

Usage

1
2
3
4
5
6
7
multi_de_pairs(summarized = NULL, paired = "unpaired",
  intercept = NULL, adjust_method = "BH", EDASeq_method = "upper",
  norm_method = "EDASeq", ruv_correct = FALSE,
  ensembl_annotate = NULL, gtf_annotate = NULL, plot_dir = NULL,
  output_voom = NULL, output_edger = NULL, output_deseq = NULL,
  output_combined = NULL, verbose = FALSE, legend = TRUE,
  label = TRUE)

Arguments

summarized

A "RangedSummarizedExperiment" object with included groups to be analysed. For format specifications see ?buildSummarized. E.g. accessible as "summarized$group". Groups are used to automate colouring of samples in unsupervised analyses. Default = NULL

paired

Are the sample paired? If "paired" a paired statistical analysis by including factors as pairs described in the "pairs" column of the "RangedSummarizedExperiment" object in the model (accessible as summarized$pairs). Options are "unpaired" or "paired". Default="unpaired"

intercept

Optional ability to set the base term for fitting the model. This is not necessary as all pairs are computed automatically. The base term, if set, must match the name of s group in "summarized$group". Default = NULL

adjust_method

Method used for multiple comparison adjustment of p-values. Options are: "holm", "hochberg", "hommel", "bonferroni", "BH", "BY", "fdr" or "none". See ?p.adjust.methods for a details description and references. Default = "BH"

EDASeq_method

Method for normalisation (applies to QC results using EDASeq and RUV when EDASeq is selected). Options are:"median","upper","full". Default = "upper"

norm_method

Methods for normalisation. Options are: "EDASeq" or "all_defaults". When "all_defaults" is selected, this will use all default normalisation methods for differential expression, EDASeq for QC, and edgeR "upperquantile" for determining RUV residuals (as per RUVSeq vignette). When "EDASeq" is selected, this will use EDASeq normalisation throughout. EDASeq normalisation method is selected using "EDASeq_method". Default = "EDASeq".

ruv_correct

Remove Unwanted Variation (RUV)? See ?RUVr for description. Currently only RUVr, which used the residuals is enabled and one factor of variation is determined. If set to TRUE and a "plot_dir" is provided, additional plots after RUV correction and the RUV residuals will be reported. Residuals are obtained through fitting a generalised linear model (GLM) using EdgeR. Residuals are then incorporated into the SummarizedExperiment object and all models for DE analysis. Options = TRUE, FALSE. Default = FALSE.

ensembl_annotate

If the dataset has been mapped to ensembl transcript identifiers, obtain additional annotation of the ensembl transcripts. A R Genome Wide Annotation object e.g. org.Mm.eg.db for mouse or org.Hs.eg.db for human must be provided. Default = NULL

gtf_annotate

Full path to a gtf file describing the transcripts. If provided will obtain gene symbols from gtf file. If a ensembl_annotate object is also provided, this will extract annotations based on the symbols extracted from the GTF file. It is recommended to provide both a gtf file and a tx_db for better annotation results. Default = NULL

plot_dir

Full path to directory for output of plots (pdf files). See ?diag_plots for more details. Default = NULL

output_voom

If you wish to output the results of the Voom analysis, provide a full path to directory for output of files. Default = NULL

output_edger

If you wish to output the results of the EdgeR analysis, provide a full path to directory for output of files. Default = NULL

output_deseq

If you wish to output the results of the DEseq2 analysis, provide a full path to directory for output of files. Default = NULL

output_combined

consensusDE will report the results of Voom, EdgeR and DEseq2 as a combined report. If you wish to output the results of the COMBINED analysis, provide a full path to directory for output of files. In addition to the combined data, it will also output the raw count and normalised data to the same directory. Default = NULL

verbose

Verbosity ON/OFF. Default=FALSE

legend

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

label

Include point labels in plots? Points are based on ID column after DE analysis from merged results. Options: TRUE, FALSE. Default = TRUE

Value

A list of all the comparisons conducted. ## See vignette for more details.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
## Load the example data set and attach - see vignette for more details
## The example below will perfrom DE analysis on all pairs of data
library(airway)
data(airway)
## Name groups of the data.
colData(airway)$group <- colData(airway)$dex
## Identify file locations
colData(airway)$file <- rownames(colData(airway))
#' ## Filter low count data:
airway_filter <- buildSummarized(summarized = airway,
                                 filter = TRUE)
## for illustration, we only use random sample of 1000 transcripts
set.seed(1234)
airway_filter <- sample(airway_filter, 1000)
## Run multi_de_pairs() with-out RUV correction
## To run with RUV correction, use ruv_correct = TRUE
all_pairs_airway <- multi_de_pairs(summarized = airway_filter,
                                   ruv_correct = FALSE,
                                   paired = "unpaired")

consensusDE documentation built on Nov. 8, 2020, 5:32 p.m.