Awesome

Grape

Grape provides an extensive pipeline for RNA-Seq analyses. It allows the creation of an automated and integrated workflow to manage and analyze RNA-Seq data.

It uses Nextflow as the execution backend. Please check Nextflow documentation for more information.

Grape has been adopted for RNA-seq integrative analysis within the IHEC consortium. Check the IHEC setup document to run the pipeline following IHEC recommendations.

Requirements

• Unix-like operationg system (Linux, MacOS, etc)
• Java 11 or later
• Docker or Singularity engine

Quickstart

1. Install Nextflow by using the following command:

   curl -s https://get.nextflow.io | bash
   

2. Make a test run:

   nextflow run guigolab/grape-nf -with-docker
   

NOTE: the very first time you execute it, it will take a few minutes to download the pipeline from this GitHub repository and the associated Docker images needed to execute the pipeline.

Pipeline software

The preferred way to run the pipeline is to use Docker or Singularity to provision the programs needed for the execution. Just use the -with-docker or -with-singularity option in the pipeline command. Pre-built Grape containers are publicly available at the Grape page in Docker Hub.

Using Singularity

Singularity is the preferred container engine for running the pipeline in an HPC environment. In order to minimize the amount of issues that could arise we recommend the use of Singularity version 3.0 or higher.

Image cache dir

The first time you run the pipeline with Singularity it will download the required images from the Docker Hub and save them in a folder inside the pipeline work dir. You can specify a different location (e.g. a centralized cache) by using the NXF_SINGULARITY_CACHEDIR environment variable or by including the following snippet in a file called nextflow.config and placing it in the current working folder of your pipeline:

singularity {
  cacheDir = "/data/singularity"
}

Please check the Singularity section in Nextflow documentation for more information.

Bind mounts

Nextflow expects that data paths are defined system wide, and your Singularity images need to be able to access these paths. Singularity allows paths that do not currently exist within the container to be created and mounted dynamically by specifying them on the command line. For this to work the user bind control option must be set to true in the Singularity config file. Nextflow support for this feature is enabled by default for the pipeline, by defining the singularity.autoMounts = true setting in the main configuration file.

Starting in version 3.0, Singularity can bind paths to non-existent mount points within the container even in the absence of the “overlay fs” feature, thus supporting architectures running legacy kernel versions (including RHEL6 vintage systems). For older versions of Singularity a kernel supporting the OverlayFS union mount filesystem is required for this functionality to be supported.

Please see here for further instructions on Singularity mounts.

Pipeline parameters

A usage message is provided and can be seen using the --help pipeline option in the command as follows:

nextflow run guigolab/grape-nf --help

--index INDEX_FILE

• specifies the path of the file containing the list of input files and the corresponding metadata (see the next section for more details).

--genome GENOME_FILE

• sets the location of the input genome FASTA file

--annotation ANNOTATION_FILE

• sets the location of the input GTF/GFF annotation file

--steps STEP[,STEP]..

• defines the pipeline steps to be performed

--paired-end

• specifies that the data is paired-end (to be used whith BAM input files)

Mapping options

--max-mismatches THRESHOLD

• set a maximum threashold for the number of allowed mismatches

--max-multimaps THRESHOLD

• set a maximum threashold for the number of allowed multiple mapped reads

--bam-sort METHOD

• set the sort method of the out BAM file

--add-xs

• add the SAM tag XS to the output BAM file (useful for using the file with tools like Cufflinks or StringTie that use the tag to know the directionality of the split maps)

Read group options

These options are used to customize the @RG header tag of the BAM files produced by the mapping step, according to the SAM specifications.

--rg-platform PLATFORM

• set the PL attribute

--rg-library LIBRARY

• set the LB attribute

--rg-center-name CENTER_NAME

• set the CN attribute

--rg-desc DESCRIPTION

• set the DS attribute

Pipeline input

The pipeline reads the paths of the FASTQ/BAM files to be processed and the corresponding metadata from a TSV file (see the --index parameter). The file must contains the following columns in order:

NOTE: Please do not use Excel/Libreoffice or similar programs to create this file. You can use this online TSV editor, also available as a VSCode extension.

NOTE: Fastq files from paired-end data will be grouped together by runID.

Here is an example from the test run:

sample1  test1   data/test1_1.fastq.gz   fastq   FqRd1
sample1  test1   data/test1_2.fastq.gz   fastq   FqRd2

Sample and id can be the same in case you don't have/know sample identifiers:

run1  run1   data/test1_1.fastq.gz   fastq   FqRd1
run1  run1   data/test1_2.fastq.gz   fastq   FqRd2

Pipeline results

The paths of the resulting output files and the corresponding metadata are stored into the pipeline.db file (TSV formatted) which sits inside the current working folder. The format of this file is the same as the index file with few more columns:

Here is an example from the test run:

sample1   test1   /path/to/results/sample1.contigs.bed    bed      Contigs                     Paired-End   MATE2_SENSE
sample1   test1   /path/to/results/sample1.isoforms.gtf   gtf      TranscriptQuantifications   Paired-End   MATE2_SENSE
sample1   test1   /path/to/results/sample1.plusRaw.bw     bigWig   PlusRawSignal               Paired-End   MATE2_SENSE
sample1   test1   /path/to/results/sample1.genes.gff      gtf      GeneQuantifications         Paired-End   MATE2_SENSE
sample1   test1   /path/to/results/test1_m4_n10.bam       bam      GenomeAlignments            Paired-End   MATE2_SENSE
sample1   test1   /path/to/results/sample1.minusRaw.bw    bigWig   MinusRawSignal              Paired-End   MATE2_SENSE

Output files

The pipeline produces several output files during the workflow execution. Many files are to be considered temporary and can be removed once the pipeline completes. The following files are the ones reported in the pipeline.db file and are to be considered as the pipeline final output.

Alignments to the reference genome

This BAM file contains information on the alignments to the reference genome. It includes all the reads from the FASTQ input. Reads that do not align to the reference are set as unmapped in the bam file. The file can be the product of several steps of the pipeline depending on the given input parameters. It is initially produced by the mapping step, then it can be the result of merging of different runs from the same experiment and finally it can run through a marking duplicates process that can eventually remove reads that are marked as duplicates.

Alignments to the reference transcriptome

This BAM file contains information on the alignments to the reference transcriptome. It is generally used only for expression abundance estimation, as input in the quantification process. The file is generally produced in the mapping process and can be the result of merging of different runs from the same experiment.

Alignments statistics

This JSON file contains alignment statistics computed with the bamstats program. It also reports RNA-Seq quality check metrics agreed within the IHEC consortium.

Signal tracks

These BigWig files (one or two, depending on the strandedness of the input data) represent the RNA-Seq signal.

Contigs

This BED file reports RNA-seq contigs computed from the pooled signal tracks.

Quantifications

These two files report abundances for genes and transcripts in the processed RNA-seq samples. The format can be either GFF or TSV depending on the tool used to perform the quantification.

Pipeline configuration

Executors

Nextflow provides different Executors run the processes on the local machine, on a computational cluster or different cloud providers without the need to change the pipeline code.

By default the local executor is used, but it can be changed by using the executor configuration scope.

For example, to run the pipeline in a computational cluster using Sun Grid Engine you can create a nextflow.config file in your current working directory with something like:

process {
    executor = 'sge'
    queue    = 'my-queue'
    penv     = 'smp'
}

Pipeline profiles

The Grape pipeline can be run using different configuration profiles. The profiles essentially allow the user to run the analyses using different tools and configurations. To specify a profile you can use the -profiles Nextflow option.

The following profiles are available at present:

The default profile is starrsem.

Run the pipeline

Here is a simple example of how you can run the pipeline:

nextflow -bg run grape-nf -r v1.1.4 --index input-files.tsv --genome refs/hg38.AXYM.fa --annotation refs/gencode.v21.annotation.AXYM.gtf --rg-platform ILLUMINA --rg-center-name CRG -resume > pipeline.log

It is strongly recommended to run one of the pipeline released versions unless you have a very good reason not to do so. This is done via the -r command line option as shown in the command above. Please see this section of Nextflow documentation for more details on this.

By default the pipeline execution will stop as far as one of the processes fails. This behaviour can be changed using the errorStrategy process directive, which can also be specified on the command line. For example, to ignore errors and keep processing you can use:

-process.errorStrategy=ignore.

It is also possible to run a subset of the pipeline steps using the option --steps. For example, the following command will only run mapping and quantification:

nextflow -bg run grape-nf --steps mapping,quantification --index input-files.tsv --genome refs/hg38.AXYM.fa --annotation refs/gencode.v21.annotation.AXYM.gtf --rg-platform ILLUMINA --rg-center-name CRG > pipeline.log

Tools versions

The pipeline can be also run natively by installing the required software on the local system or by using Environment Modules.

The versions of the tools that have been tested so far with the standard pipeline profile are the following:

• bamstats v0.3.4
• bedtools v2.19.1
• KentUtils v308
• RSEM v1.2.21
• RSeQC v2.6.4
• sambamba v0.7.1
• samtools v1.3
• STAR v2.4.0j