DESeq2 generalities

This module allows to perform the normalization and the differential expression analysis on expression step results. This module is based on the DESeq2 package of Bioconductor.

  • Name: DESeq2

  • Input port:
    • input: expression files in TSV format (format: expression_results_tsv)

  • Output:
    • Graphs for the normalization:
      • Unpooled sample clustering.
      • Unpooled PCA.
      • null counts barplot: percentage of genes with null counts per sample.
      • unpooled counts barplot: number of counts per sample.
      • unpooled counts boxplot.
      • pooled counts barplot: number of counts per sample (After RepTech pooling).
      • pooled counts boxplot.
      • pooled and normalized dendrogram.
      • pooled and normalized PCA.
      • pooled and normalized boxplot.
      • most expressed features plot: percentage of reads on the gene with the maximum number of reads.
    • Graphs for the differential analysis:
      • dispersion plot.
      • p-value distribution plot.
      • adjusted p-value distribution plot.
      • MA plot.
      • Number of differentially expressed genes plot.
      • volcano plot (easy-contrasts-DESeq2 v2 only).
      • Heatmap (easy-contrasts-DESeq2 v2 only).
    • TSV files:
      • raw counts matrix.
      • pooled counts matrix.
      • normalized counts matrix.
      • diffana matrix: one for each comparison performed during the differential expression analysis.
    • Other files:
      • HTML report file (easy-contrasts-DESeq2 v2 only).
      • RDS file for import statistical analysis in R (easy-contrasts-DESeq2 v2 only).

Workflow parameters for DESeq2

  • Parameters :
    Parameter Type Description Default value easy-contrasts-DESeq2 compatibility
    easy.contrasts.version integer Version of Easy-contrasts-DESeq2 to use. 2 N/A
    norm.fig boolean If true, generate the figures of the normalization. true v1 only
    diffana.fig boolean If true, generate the figures of the differential analysis. true v1 only
    norm.diffana boolean If set to false, skip the normalization and the differential analysis. This option is used to produce the contrast matrix without doing the normalization and the differential analysis. true v1 only
    diffana boolean If set to false, skip the differential analysis. true all versions
    size.factors.type string Determination of the size factors type to use during the differential analysis (value: ratio or iterate). ratio all versions
    fit.type string Determination of the fit type for the distance estimation to use during the differential analysis (value: parametric, local or mean). parametric all versions
    statistical.test string Statistical test to be used during the differential analysis (value: Wald or LRT). Wald all versions
    adjusted.p.value.threshold float Adjusted p-value threshold. 0.05 v2 only
    log2.fc.threshold float Log2 fold change threshold. 2.0 v2 only
    disable.nested.models boolean Disable nested models. false v2 only
    use.additional.annotation.file boolean Set if an additional annotation file must be used. true v2 only
    r.execution.mode string The R execution mode. The available mode values are: process, rserve and docker. process all versions
    rserve.servername string The Rserve server name to use in rserve execution mode not set all versions
    docker.image string The Docker image to use in Docker execution mode. bioconductor/release_sequencing:3.1 all versions
    logo.url string Logo URL on the HTML report. Eoulsan logo v2 only
    author.name string Author name. Eoulsan v2 only
    author.email boolean Author email. eoulsan@biologie.ens.fr v2 only
    save.rds boolean Save or not RDS object of the analysis. false v2 only
    save.r.scripts boolean Save or not R scripts. false all versions

Required R packages installation

Eoulsan differential expression analysis module use easy-contrasts-DESeq2 R scripts to perform statistical analysis. This R scripts are bundled with Eoulsan, however they need R and some R packages to work. Requirements for easy-contrasts-DESeq2 are described on the easy-contrasts-DESeq2 project page on GitHub. You can install DESeq2 and required R packages on your computer, on a Rserve server, or you can use the Docker image containing all the packages needed to run DESeq2:

How to use DESeq2?

DESeq2 can be used in 2 different modes: classic mode or contrast mode.

Classic mode

The classic mode performs the differential analysis on the "Condition" column of the design file. During the differential expression analysis each condition is compared to the others. The "Reference" column can be used to specify references for the comparison.

  • In the design file: To design an experiment with a classical mode differential expression analysis you can use the following example in the design file: 
    Exp.1.name=exp1
Exp.1.skip=false
Exp.1.reference=WT-day1

    The keys "reference" is optional and can be replaced by a "Reference" column in the design file.

Contrast mode

The contrast mode performs the differential analysis using contrast vector.

  • In the design file: To design an experiment with a contrast mode differential analysis you can use the following example in the design file: 
    Exp.2.name=exp2
Exp.2.skip=false
Exp.2.contrast=true
Exp.2.buildContrast=true
Exp.2.model=~type+day+type:day
Exp.2.comparisons=WT1_vs_KO1:typeWT%dayday1_vs_typeKO%dayday1;\
WT2_vs_KO2:typeWT%dayday2_vs_typeKO%dayday2

    • You need to set the options contrast and buildContrast at true.
    • You need to set the DESeq2 model including the the column needed for the contrast.
    • You need the comparisons options where you specify the comparison to be done according the contrasts. The contrast vectors will be generated from this option. See the Design file v2 section for more information on this option.