The rseqR package is designed to perform differential analysis of RNA-seq data. It provides tools to:
- perform quality control of raw reads in fastq format and generate HTML reports (using FASTQC and MULTIQC)
- determine trancript and gene-level read count estimates (using Salmon)
- run differential analysis using DESeq2, edgeR, and Limma-voom and to export the results into easily readable tab-delimited files
- perform pathway analysis of differentially expressed genes
The rseqR package does not attempt to replace the tools it uses, but instead provides an automated environment in which to use them.
The website for rseqR may be found here.
The workflow used by rseqR requires the installation of several command-line tools: Fastqc (v0.11.7), MultiQC (v1.6), and Salmon (v0.10.2). Installation instructions for these may be found at their respective websites, but a guide is given below for convenience.
If using a Unix-based system, open up a terminal and follow the commands as is. If using the windows subsystem for linux (WSL) on Windows 10 then WSL must first be set up as detailed here. Once WSL is up and running, then the tools may be installed as on any Unix-based system.
It is assumed that root priviliges are available (if necessary) - if not, then a system administrator may need to install these for you.
cd ~
wget https://www.bioinformatics.babraham.ac.uk/projects/fastqc/fastqc_v0.11.7.zip
unzip fastqc_v0.11.7.zip
cd FastQC
chmod 755 fastqc
# If using WSL
cd /usr/local/bin
sudo ln -s ~/Fastqc/fastqc .
# If using other Unix
cd ~/bin
ln -s ~/Fastqc/fastqc .
To run MultiQC ensure there is a working python distribution.
pip install multiqc
cd ~
wget https://github.com/COMBINE-lab/salmon/releases/download/v0.11.2/salmon-0.11.2-linux_x86_64.tar.gz
tar -xvzf salmon-0.11.2-linux_x86_64.tar.gz
# If using WSL
cd /usr/local/bin
sudo ln -s ~/salmon-0.11.2-linux_x86_64/bin/salmon .
# If using other Unix
cd ~/bin
ln -s ~/salmon-0.11.2-linux_x86_64/bin/salmon .
There are a number of dependencies to the rseqR package detailed in the
DESCRIPTION
file. These will all be installed automatically during the
installation of rseqR.
Once all dependencies are installed, then rseqR may be installed as follows:
devtools::install_github("anilchalisey/rseqr", build_vignettes = TRUE)
The entire workflow of rseqR may be run by a call to a single function
run_dea()
, as shown
below.
result <- run_dea(sample.info = `system.file("extdata", "paired-example.txt", pkg = "rseqr")`,
reference = c("WT", "KO1", "KO2"), species = "human", output.dir = "results-dea",
threads = NULL, index.dir = NULL)
The arguments to the run_dea()
function are:
This is the path to a tab-delimited file with at least the columns:
condition
: treatment or condition labels on the basis of which the differential analysis will be performed (e.g. “CONTROL”, “TREATED”; or “WILD-TYPE”, “KNOCK-OUT1”, “KNOCK-OUT2”).sample
: names of the samples - these labels will be used in plots and as sample column headers in the output.file1
: absolute or relative path to fastq or salmon quant.sf files.
Optional columns include:
file2
: if fastq files and PE reads, then this column should also be present, specifying the absolute or relative paths to the second pair of fastq reads.
#batch
: if a batch effect is to be included in the design, then this should be identified under this column (e.g. litter number or sequencing run).
If starting from fastq files, then the workflow will include QC of raw reads, and read count estimation with salmon. If a salmon index has not been identified then an index will be built automatically prior to read count estimation. If starting from quant.sf files, then the analysis will start directly at the differential analysis stage.
Dummy examples of the tab-delimited files accepted by rseqR may be found
at system.file("extdata", "paired-example.txt", pkg = "rseqr")
,
system.file("extdata", "sinle-example.txt", pkg = "rseqr")
, and
system.file("extdata", "quants-example.txt", pkg = "rseqr")
.
The order in which the condition labels should be evaluated when performing differential analysis. For example, c(“A”, “B”, “C”, “D”) would mean “A” is the reference condition to which “B”, “C” and “D” are compared; in addition, “C” and “D” will be compared to “B”, and “D” will be compared to “C”. If then the comparisons will be arranged alphabetically.
May be one of “human” or “mouse” - other options will be added in later versions.
Directory to which results should be saved.
Number of threads to be used when parallelisation is possible. If NULL
then one less than the maximum numbre of threads available will be used.
The path to the salmon index.
Other arguments are also possible, and details for these may be found in the manual, but for most users, the default settings are satisfactory.
Once complete, all the results will be saved in the specified output directory with the following structure:
#> levelName
#> 1 results
#> 2 ¦--DE_results.rda
#> 3 ¦--DESeq2
#> 4 ¦--edgeR
#> 5 ¦--fastqc
#> 6 ¦--index
#> 7 ¦--Limma-voom
#> 8 ¦--metadata.rda
#> 9 ¦--multiqc
#> 10 ¦--salmon
#> 11 ¦--upset_KO1-WT.png
#> 12 ¦--upset_KO2-WT.png
#> 13 °--upset_KO2-KO1.png
The directories DESeq2
, edgeR
, fastqc
, Limma-voom
, and multiqc
contain the results of the respective analyses including relevant plots.
Within the salmon
directory one will find individual directories for
the quantification results of each sample. The index
directory
contains the salmon index for human or mouse if it had not been created
and specified beforehand. DE-results.rda
contains the results of the
differential expression analysis and comprises a list of three objects:
DE_limma
, DE_edger
, and DE_deseq2
. The metadata.rda
object
contains the metadata for the samples including the count data for each
sample; this may be used to re-run the analysis from an intermediate
step.
If there is a need to reanalyse only a portion of the data, then this
may be done using the metadata.rda
object, which contains all the
necessary information. For example, it is possible to run each of the
differential analysis tools separately using the commands
limma_voom_analysis()
, edger_analysis()
, or deseq2_analysis()
. For
example:
load("results/metadata.rda") # this will load an object called 'metadata' into the environment
DE_limma <- limma_voom_analysis(metadata = metadata, species = "human")
Gene ontology and KEGG pathway analysis may be performed using the
gage
package. This is performed by reading in the differential
analysis results written out to file and submitting this to the
pathway_analysis()
function. In this example, we use the edgeR
results.
DEGs <- read.csv("results/edgeR/KO1-WT/edgeR_differential_expression.txt")
pa1_deseq2 <- pathway_analysis(DEGs, species = "human")
This may be repeated as necessary for the other comparisons.
The rseqR package has been developed in the Chris O’Callaghan Group at the Centre for Cellular and Molecular Biology, University of Oxford by Anil Chalisey and Chris O’Callaghan.