What it does

The snakePipes RNA-seq workflow allows users to process their single or paired-end RNA-Seq fastq files upto the point of gene/transcript-counts and differential expression. It also allows full allele-specific RNA-Seq analysis (up to allele-specific differential expression) using the allelic-mapping mode.


Input requirements

The only requirement is a directory of gzipped fastq files. Files could be single or paired end, and the read extensions could be modified using the keys in the defaults.yaml file below.

Configuration file

There is a configuration file in snakePipes/workflows/RNA-seq/defaults.yaml:

## General/Snakemake parameters, only used/set by wrapper or in Snakemake cmdl, but not in Snakefile
pipeline: rna-seq
local: False
maxJobs: 5
## directory with fastq files
## preconfigured target genomes (mm9,mm10,dm3,...) , see /path/to/snakemake_workflows/shared/organisms/
## Value can be also path to your own genome config file!
## FASTQ file extension (default: ".fastq.gz")
ext: '.fastq.gz'
## paired-end read name extension (default: ["_R1", "_R2"])
reads: ["_R1","_R2"]
## Number of reads to downsample from each FASTQ file
## Options for trimming
trim: False
trimmer: cutadapt
## further options
mode: alignment-free,deepTools_qc
bwBinSize: 25
fastqc: False
featureCountsOptions: -C -Q 10 --primary
fragmentLength: 200
libraryType: 2
salmonIndexOptions: --type quasi -k 31
dnaContam: False
## supported mappers: STAR HISAT2
aligner: STAR
## N.B., setting --outBAMsortingBinsN too high can result in cryptic errors
verbose: False
plotFormat: png
# for allele-spcific mapping

Apart from the common workflow options (see Running snakePipes), the following parameters are useful to consider:

  • aligner: You can choose either STAR or HISAT2. While HISAT2 requires less memory than STAR, we keep STAR as the default aligner due to its superior accuracy (see this paper). Make sure that --alignerOptions matches this.

  • alignerOptions: Options to pass on to your chosen aligner. Note that library type and junction definitions doesn't have to be passed to the aligners as options, as they are handeled either automatically, or via other parameters.

  • featureCountsOptions: Options to pass to featureCounts (in case the alignment or allelic-mapping mode is used). Note that the paired-end information is automatically passed to featurecounts via the workflow, and the summerization is always performed at gene level, since the workflow assumes that featurecounts output is being used for gene-level differential expression analysis. If you wish to perform a transcript-level DE analysis, please choose the mode alignment-free.

  • filterGTF: Options you can pass on to filter the original GTF file. This is useful in case you want to filter certain kind of transcripts (such as pseudogenes) before running the counts/DE analysis.

  • libraryType: The default library-type is suitable for most RNAseq protocols (using Illumina Tru-Seq). Change this option in case you have a different strandednes.

  • salmonIndexOptions: In the alignment-free mode (see below), this option allows you to change the type of index created by salmon. New users can leave it to default.

  • dnaContam: Enable this to test for possible DNA contamination in your RNA-seq samples. DNA contamination is quantified as the fraction of reads falling into intronic and intergenic regions, compared to those falling into exons. Enabling this option would produce a directory called GenomicContamination with .tsv files containing this information.

  • plotFormat: You can switch the type of plot produced by all deeptools modules using this option. Possible choices : png, pdf, svg, eps, plotly

  • SNPFile: For the allelic-mapping mode. The SNPFile is the file produced by SNPsplit after creating a dual-hybrid genome. The file has the suffix .vcf.

  • NMaskedIndex: For the allelic-mapping mode. The NMaskedIndex refers to the basename of the index file created using STAR, on the SNPsplit output.


SNPFile and NMaskedIndex file could be specified in case you already have this and would like to re-run the analysis on new data. In case you are running the allele-specific analysis for the first time, you would need a VCF file and the name of the two strains. In this case the SNPFile as well as the NMaskedIndex files would be automatically created by the workflow using SNPsplit.

Differential expression

Like the other workflows, differential expression can be performed using the --sampleSheet option and supplying a sample sheet like that below:

name    condition
sample1      eworo
sample2      eworo
SRR7013047      eworo
SRR7013048      OreR
SRR7013049      OreR
SRR7013050      OreR


The first entry defines which group of samples are control. This way, the order of comparison and likewise the sign of values can be changed. The DE analysis might fail if your sample names begin with a number. So watch out for that!

Complex designs with blocking factors

If the user provides additional columns between 'name' and 'condition' in the sample sheet, the variables stored there will be used as blocking factors in the order they appear in the sample sheet. Eg. if the first line of your sample sheet looks like 'name batch condition', this will translate into a formula batch + condition. 'condition' has to be the final column and it will be used for any statistical inference.

Analysis modes

Following analysis (modes) are possible using the RNA-seq workflow:


In this mode, the pipeline uses one of the selected aligners to create BAM files, followed by gene-level quantification using featureCounts. Gene-level differential expression analysis is then performed using DESeq2.


allelic-mapping mode follows a similar process as the "mapping" mode, however the alignment performed on an allele-masked genome, followed by allele-specific splitting of mapped files. Gene-level quantification is performed for each allele using featureCounts. Allele-specific, gene-level differential expression analysis is then performed using DESeq2.


allelic-mapping mode is mutually exclusive with mapping mode


In this mode, the pipeline uses salmon to perform transcript-level expression quantification. This mode performs both transcript-level differential expression (using Sleuth), and gene-level differential expression (using wasabi, followed by DESeq2).


The salmon index is recreated each time in alignment-free mode. This is done to facilitate changing how the GTF file is filtered, which necessitates reindexing.


The pipeline provides multiple quality controls through deepTools, which can be triggered using the deepTools_qc mode. It's a very useful add-on with any of the other modes.


Since most deeptools functions require an aligned (BAM) file, the deepTools_qc mode will additionally perform the alignment of the fastq files. However this would not interfere with operations of the other modes.

Understanding the outputs

Assuming the pipline was run with --mode 'alignment-free,alignment,deepTools_qc' on a set of FASTQ files, the structure of the output directory would look like this (files are shown only for one sample)

├── Annotation
│   ├── filter_command.txt
│   ├── genes.annotated.bed
│   ├── genes.filtered.bed
│   ├── genes.filtered.fa
│   ├── genes.filtered.gtf
│   ├── genes.filtered.symbol
│   ├── genes.filtered.t2g
├── bamCoverage
│   ├── logs
│   ├── sample1.coverage.bw
│   ├── sample1.RPKM.bw
│   ├── sample1.uniqueMappings.fwd.bw
│   └── sample1.uniqueMappings.rev.bw
├── cluster_logs
├── deepTools_qc
│   ├── bamPEFragmentSize
│   │   ├── fragmentSize.metric.tsv
│   │   └── fragmentSizes.png
│   ├── estimateReadFiltering
│   │   └── sample1_filtering_estimation.txt
│   ├── logs
│   ├── multiBigwigSummary
│   ├── plotCorrelation
│   │   ├── correlation.pearson.bed_coverage.heatmap.png
│   │   ├── correlation.pearson.bed_coverage.tsv
│   │   ├── correlation.spearman.bed_coverage.heatmap.png
│   │   └── correlation.spearman.bed_coverage.tsv
│   ├── plotEnrichment
│   │   ├── plotEnrichment.png
│   │   └── plotEnrichment.tsv
│   └── plotPCA
│       ├── PCA.bed_coverage.png
│       └── PCA.bed_coverage.tsv
├── DESeq2_Salmon_sampleSheet
│   ├── DESeq2_Salmon.err
│   ├── DESeq2_Salmon.out
│   ├── citations.bib
│   ├── DESeq2_report_files
│   ├── DESeq2_report.html
│   ├── DESeq2_report.Rmd
│   ├── DESeq2.session_info.txt
│   ├── DEseq_basic_counts_DESeq2.normalized.tsv
│   ├── DEseq_basic_DEresults.tsv
│   └── DEseq_basic_DESeq.Rdata
├── DESeq2_sampleSheet
│   ├── DESeq2.err
│   ├── DESeq2.out
│   ├── citations.bib
│   ├── DESeq2_report_files
│   ├── DESeq2_report.html
│   ├── DESeq2_report.Rmd
│   ├── DESeq2.session_info.txt
│   ├── DEseq_basic_counts_DESeq2.normalized.tsv
│   ├── DEseq_basic_DEresults.tsv
│   └── DEseq_basic_DESeq.Rdata
│   ├── sample1_R1.fastq.gz
│   └── sample1_R2.fastq.gz
├── featureCounts
│   ├── counts.tsv
│   ├── sample1.counts.txt
│   ├── sample1.counts.txt.summary
│   ├── sample1.err
│   ├── sample1.out
├── multiQC
│   ├── multiqc_data
│   ├── multiQC.err
│   ├── multiQC.out
│   └── multiqc_report.html
├── QC_report
│   └── QC_report_all.tsv
├── RNA-seq.cluster_config.yaml
├── RNA-seq.config.yaml
├── RNA-seq_organism.yaml
├── RNA-seq_pipeline.pdf
├── RNA-seq_run-1.log
├── Salmon
│   ├── counts.genes.tsv
│   ├── counts.tsv
│   ├── Salmon_counts.log
│   ├── Salmon_genes_counts.log
│   ├── Salmon_genes_TPM.log
│   ├── SalmonIndex
│   ├── Salmon_TPM.log
│   ├── sample1
│   ├── sample1.quant.genes.sf
│   ├── sample1.quant.sf
│   ├── TPM.genes.tsv
│   └── TPM.tsv
├── sleuth_Salmon_sampleSheet
│   ├── logs
│   ├── MA-plot.pdf
│   ├── sleuth_live.R
│   ├── so.rds
│   └── Wald-test.results.tsv
└── STAR
        ├── logs
    ├── sample1
    ├── sample1.bam
    └── sample1.bam.bai


The _sampleSheet suffix for the DESeq2_sampleSheet and sleuth_Salmon_sampleSheet is drawn from the name of the sample sheet you use. So if you instead named the sample sheet mySampleSheet.txt then the folders would be named DESeq2_mySampleSheet and sleuth_Salmon_mySampleSheet. This facilitates using multiple sample sheets.

Apart from the common module outputs (see Running snakePipes), the workflow would produce the following folders:

  • Annotation: This folder would contain the GTF and BED files used for analysis. In case the file has been filtered using the --filterGTGTFF option (see Configuration file), this would contain the filtered files.

  • STAR/HISAT2: (not produced in mode alignment-free) This would contain the output of RNA-alignment by STAR or HISAT2 (indexed BAM files).

  • featureCounts: (not produced in mode alignment-free) This would contain the gene-level counts (output of featureCounts), on the filtered GTF files, that can be used for differential expression analysis.

  • bamCoverage: (not produced in mode alignment-free) This would contain the bigWigs produced by deepTools bamCoverage . Files with suffix .coverage.bw are raw coverage files, while the files with suffix RPKM.bw are RPKM-normalized coverage files.

  • deepTools_QC: (produced in the mode deepTools_QC) This contains the quality checks specific for RNA-seq, performed via deepTools. The output folders are names after various deepTools functions and the outputs are explained under deepTools documentation. In short, they show the insert size distribution(bamPEFragmentSize), mapping statistics (estimateReadFiltering), sample-to-sample correlations and PCA (multiBigwigSummary, plotCorrelation, plotPCA), and read enrichment on various genic features (plotEnrichment)

  • DESeq2_[sampleSheet]/DESeq2_Salmon_[sampleSheet]: (produced in the modes alignment or alignment-free, only if a sample-sheet is provided.) The folder contains the HTML result report DESeq2_report.html, the annotated output file from DESeq2 (DEseq_basic_DEresults.tsv) and normalized counts for all samples, produced via DEseq2 (DEseq_basic_counts_DESeq2.normalized.tsv) as well as an Rdata file (DEseq_basic_DESeq.Rdata) with the R objects dds <- DESeq2::DESeq(dds) and ddr <- DDESeq2::results(dds,alpha = fdr). DESeq2_[sampleSheet] uses gene counts from featureCounts/counts.tsv, whereas DESeq2_Salmon_[sampleSheet] uses transcript counts from Salmon/counts.tsv that are merged via tximport in R.

  • Salmon: (produced in mode alignment-free) This folder contains transcript-level (counts.tsv)and gene-level (counts.genes.tsv) counts estimated by the tool Salmon .

  • sleuth_Salmon_[sampleSheet] (produced in mode alignment-free, only if a sample-sheet is provided) This folder contains a transcript-level differential expression output produced using the tool Sleuth .

Command line options

MPI-IE workflow for RNA mapping and analysis

usage example:

RNA-seq -i input-dir -o output-dir mm10

usage: RNA-seq -i INDIR -o OUTDIR [-h] [-v] [--ext EXT] [--reads READS READS]
               [-c CONFIGFILE] [--clusterConfigFile CLUSTERCONFIGFILE]
               [-j INT] [--local] [--keepTemp]
               [--snakemakeOptions SNAKEMAKEOPTIONS] [--DAG] [--version]
               [--emailAddress EMAILADDRESS] [--smtpServer SMTPSERVER]
               [--smtpPort SMTPPORT] [--onlySSL] [--emailSender EMAILSENDER]
               [--smtpUsername SMTPUSERNAME] [--smtpPassword SMTPPASSWORD]
               [--VCFfile VCFFILE] [--strains STRAINS] [--SNPfile SNPFILE]
               [--NMaskedIndex NMASKEDINDEX] [-m MODE] [--downsample INT]
               [--trim] [--trimmer {cutadapt,trimgalore,fastp}]
               [--trimmerOptions TRIMMEROPTIONS] [--fastqc] [--bcExtract]
               [--bcPattern BCPATTERN] [--UMIDedup]
               [--UMIDedupSep UMIDEDUPSEP] [--UMIDedupOpts UMIDEDUPOPTS]
               [--bwBinSize BWBINSIZE] [--plotFormat STR]
               [--libraryType LIBRARYTYPE] [--aligner ALIGNER]
               [--alignerOptions ALIGNEROPTIONS]
               [--salmonIndexOptions SALMONINDEXOPTIONS]
               [--featureCountsOptions FEATURECOUNTSOPTIONS]
               [--filterGTF FILTERGTF] [--sampleSheet SAMPLESHEET]
               [--dnaContam] [--fromBAM] [--singleEnd]

Positional Arguments


Genome acronym of the target organism. Either a yaml file or one of: mm10_gencodeM13, dm6, GRCz10, dm3, mm10, mm9, hs37d5, hg38, SchizoSPombe_ASM294v2

Required Arguments

-i, --input-dir

input directory containing the FASTQ files, either paired-end OR single-end data

-o, --output-dir

output directory

General Arguments

-v, --verbose

verbose output (default: 'False')


Suffix used by input fastq files (default: '".fastq.gz"').


Suffix used to denote reads 1 and 2 for paired-end data. This should typically be either '_1' '_2' or '_R1' '_R2' (default: '['_R1', '_R2']). Note that you should NOT separate the values by a comma (use a space) or enclose them in brackets.

-c, --configFile

configuration file: config.yaml (default: 'None')


configuration file for cluster usage. In absence, the default options specified in defaults.yaml and workflows/[workflow]/cluster.yaml would be selected (default: 'None')

-j, --jobs

maximum number of concurrently submitted Slurm jobs / cores if workflow is run locally (default: '5')


run workflow locally; default: jobs are submitted to Slurm queue (default: 'False')


Prevent snakemake from removing files marked as being temporary (typically intermediate files that are rarely needed by end users). This is mostly useful for debugging problems.


Snakemake options to be passed directly to snakemake, e.g. use --snakemakeOptions='--dryrun --rerun-incomplete --unlock --forceall'. WARNING! ONLY EXPERT USERS SHOULD CHANGE THIS! THE DEFAULT VALUE WILL BE APPENDED RATHER THAN OVERWRITTEN! (default: '['--use-conda']')


If specified, a file ending in _pipeline.pdf is produced in the output directory that shows the rules used and their relationship to each other.


show program's version number and exit

Email Arguments


If specified, send an email upon completion to the given email address


If specified, the email server to use.


The port on the SMTP server to connect to. A value of 0 specifies the default port.


The SMTP server requires an SSL connection from the beginning.


The address of the email sender. If not specified, it will be the address indicated by --emailAddress


If your SMTP server requires authentication, this is the username to use.


If your SMTP server requires authentication, this is the password to use.

Allele-specific mapping arguments


VCF file to create N-masked genomes (default: 'None')


Name or ID of SNP strains separated by comma (default: 'None')


File containing SNP locations (default: 'None')


N-masked index of the reference genome (default: 'None')


-m, --mode

workflow running modes (available: 'alignment-free, alignment, allelic-mapping, deepTools_qc') (default: '"alignment,deepTools_qc"')


Downsample the given number of reads randomly from of each FASTQ file (default: 'False')


Activate fastq read trimming. If activated, Illumina adaptors are trimmed by default. Additional parameters can be specified under --trimmerOptions. (default: 'False')


Possible choices: cutadapt, trimgalore, fastp

Trimming program to use: Cutadapt, TrimGalore, or fastp. Note that if you change this you may need to change --trimmerOptions to match! (default: '"cutadapt"')


Additional option string for trimming program of choice. (default: 'None')


Run FastQC read quality control (default: 'False')


To extract umi barcode from fastq file via UMI-tools and add it to the read name (default: 'False')


The pattern to be considered for the barcode. 'N' = UMI position (required) 'C' = barcode position (optional) (default: '')


Deduplicate bam file based on UMIs via umi_tools dedup that are present in the read name. (default: 'False')


umi separation character that will be passed to umi_tools.(default: '"_"')


Additional options that will be passed to umi_tools.(default: '')


Bin size of output files in bigWig format (default: '25')


Possible choices: png, pdf, None

Format of the output plots from deepTools. Select 'none' for no plots (default: '"png"')


user provided library type strand specificity. featurCounts style: 0, 1, 2 (Illumina TruSeq); default: '2')


Program used for mapping: STAR or HISAT2 (default: '"STAR"'). If you change this, please change --alignerOptions to match.


STAR or hisat2 option string, e.g.: '--twopassMode Basic' (default: 'None')


Salmon index options, e.g. '--type fmd' (default: '"--type quasi -k 31"')


featureCounts option string. The options '-p -B' are always used for paired-end data (default: '"-C -Q 10 --primary"')


filter annotation GTF by grep for use with Salmon, e.g. use --filterGTF='-v pseudogene'; default: 'None')


Information on samples (required for DE analysis); see 'docs/content/sampleSheet.example.tsv' for example. The column names in the tsv files are 'name' and 'condition'. The first entry defines which group of samples are control. This way, the order of comparison and likewise the sign of values can be changed. The DE analysis might fail if your sample names begin with a number. So watch out for that! (default: 'None')


Returns a plot which presents the proportion of the intergenic reads (default: 'False')


Input folder with bam files. If provided, the analysis will start from this point. If bam files contain single ends, please specify --singleEnd additionally.


input data is single-end, not paired-end. This is only used if --fromBAM is specified.

code @ github.