ribomethseq-nf is a nextflow pipeline dedicated to RiboMethSeq data processing. It generates quality control data and counts, which can be directly used for further analyses using the rRMSAnalyzer package.
- 1.0 : first release
Nextflow (21.04 or later) is required to run this pipeline.
Warning Older Nextflow versions before 20.10.0 will likely fail to run RiboMethSeq-nf.
The following software program/packages are also required to run this pipeline.
- bowtie2
- samtools [>=1.16.1]
- bedtools
- Trimmomatic
- FastQC
- MultiQC
- pandoc
- R and the following libraries :
- ade4
- dplyr
- tidyr
- tibble
- rmarkdown
- pheatmap
If you do not want to install them manually, you can use either docker/singularity or conda to build the required environment (see Installation section)
Note: Earlier version of samtools may work as well but will retain a few more
SAM records due to an issue in expression handling in samtools view prior to
version 1.15 (See this issue).
In our case this is related to this filtering expression -d 'NM' -e '![XS]'given
to samtools. This is expected to be really minor though.
The workflow is currently designed to process RiboMethSeq data generated in
single-end mode on Illumina sequencers. For each detected fastq file (one per
sample), quality control (FastQC) and reads trimming (Trimmomatic) steps are
launched. Trimmed reads are then aligned on rRNA reference sequences (human or mouse,
included in the pipeline) with Bowtie2 in end-to-end mode with --sensitive -L 17 parameters.
Aligned files are then coordinate sorted and filtered with samtools to obtain uniquely mapped reads.
From these, we generate count (coverage) files using bedtools. A MultiQC report
gathering metrics from FastQC, Trimmomatic and Bowtie2 is provided as well as an
HTML report with custom QC metrics computed on the full dataset (all detected
fastq files).
To have the workflow installed you simply need to clone this repository.
git clone https://github.com/RibosomeCRCL/riboMethseq-nfPerhaps the easiest way to have a proper environment to run the pipeline is to
use conda. Two YAML files are provided : docker/rms-processing.yml and
docker/rms-report.yml corresponding to two distinct environments. The former
contains requirements for the processing part of the pipeline (samtools, bowtie2, ...)
and the latter is dedicated to the report generation. We chose to split the two
environment because solving the full conda environment could take a really long time.
You can either directly use the conda profile provided in nextflow.config
and the workflow will automatically build the environment at workflow
initiation or you can build it in advance (recommended) with for instance the
following command:
cd docker
conda env create -f rms-processing.yml
conda env create -f rms-report.ymlIf you create the environment prior to running the workflow, you will then need
to adapt the process.conda directives from the conda profile in nextflow.config
as shown below :
conda {
includeConfig "nf-config/exec.config"
conda.enabled = true
process {
withName: 'fastqc|trim|bowtie2|filter|multiqc|counts' {
conda = "/path/to/your/conda/envs/rms-processing"
}
withName: 'split|report' {
conda = "/path/to/your/conda/envs/rms-report"
}
}
}
A docker image can also be built using the provided docker/Dockerfile.prod. The
image will be built upon a debian 11 (bullseye) base image.
To build the docker image :
cd docker
docker build -t ribomethseq-nf:1.0 -f Dockerfile.prod .If you need to mount specific path(s) of your infrastructure, adapt the docker
profile in the configuration file (process.containerOptions).
More information on docker image content in the docker README.
There is no proper singularity recipe provided at the moment, but in the meantime you can convert the Docker image to a singularity image.
# first, create an archive of the docker image
docker save ribomethseq-nf:1.0 | gzip > ribomethseq-nf_1.0.tar.gz
# (... somewhere else ...)
# second, build a sif image from the archive
singularity build [--sandbox] ribomethseq-nf_1.0.sif docker-archive://ribomethseq-nf_1.0.tar.gzIf problems occur with the loop device, the option --sandbox can be used to build the singularity image.
The resulting ribomethseq-nf_1.0.sif must be in the directory specifed by singularity.cacheDir in the configuration file.
If you need to mount specific path(s) of your infrastructure, adapt the Singularity
profile in the configuration file (process.containerOptions).
By default in nf-config/exec.config, the executor is set to slurm. You may
need to adapt it to your own computing infrastructure (e.g. pbs, LSF, or other)
You can also set params.scheduler = 'local' if you plan to run locally on your
computer but then pay attention to control the queue size with --qsize (set to
20 by default) to avoid launching to many jobs depending on your hardware
capabilities.
Some tests are provided in the tests directory. Once you have set up your
software environment :
cd tests
# Test with tools installed in PATH
make test
# Test with conda profile
make test-conda
# Test with docker profile on local machine
make test-docker SCHEDULER=local
# Test with singularity profile on HPC with pbs
make test-singularity SCHEDULER=pbs
# Test with your own created profile (e.g. `custom_profile`)
make test PROFILE=custom_profile
# Clean the tests directory
make cleanLet's suppose that you have cloned the repository in the following directory:
/path/to/ribomethseq-nf
To run the workflow, you will need to specify both your execution environment and your species (human or mouse) of interest. This is done here by combining nextflow profiles.
human and mouse profiles are already available. They gather for each species
all the required reference data and the bowtie indexes to save you some time.
nextflow run /path/to/ribomethseq-nf -profile human,docker \
--fqdir $FastqDir \
--outdir $OutDirnextflow run /path/to/ribomethseq-nf -profile mouse,conda \
--fqdir $FastqDir \
--outdir $OutDirSee Input parameters section for a description of all available parameters.
Easiest way to get you started for the non-bioinformatician
- You just need nextflow and conda available on your system.
- Find out what is your job scheduler. e.g.
pbs
Then:
nextflow RibosomeCRCL/ribomethseq-nf -profile conda,human --scheduler 'pbs' --qsize 10 --fqdir '/path/to/fastq/files'
This will automatically retrieve the nextflow pipeline from GitHub, build the required conda environment and finally process your data. Output files will be located in current directory (default).
The following reference rRNA are used :
- https://www.ncbi.nlm.nih.gov/nuccore/NR_046235 (18S, 28S and 5.8S)
- https://www.ncbi.nlm.nih.gov/nuccore/NR_023363.1 (5S)
- https://www.ncbi.nlm.nih.gov/nuccore/NR_030686.1 (5S)
- https://www.ncbi.nlm.nih.gov/nuccore/NR_003280.2 (5.8S)
- https://www.ncbi.nlm.nih.gov/nuccore/NR_003278.3 (18S)
- https://www.ncbi.nlm.nih.gov/nuccore/NR_003279.1 (28S)
The associated fasta sequences for human and mouse organisms are stored in the
data/fasta directory. It is possible to add other species, but it will be
necessary to add a new profile in nextflow.config file for them.
Precomputed bowtie2 indexes are also already provided for both human and mouse
in the folder data/bowtie. If you need to use your own index, you can specify
it through the --bowtie_index parameter.
+++++++++++++++++++++++++
+ ribomethseq-nf help +
+++++++++++++++++++++++++
--fqdir DIR Fastq files location Required
--fastq_pattern STR Pattern for fastq file selection Optional (.fastq.gz)
--adapters FILE (Trimmomatic) Path to illumina adapters Optional ($baseDir/data/adapters/TruSeq3-SE.fa)
--leading INT (Trimmomatic) LEADING parameter Optional (30)
--trailing INT (Trimmomatic) TRAILING parameter Optional (30)
--slidingwindow STR (Trimmomatic) SLIDINGWINDOW parameter Optional (4:15)
--avgqual INT (Trimmomatic) AVGQUAL parameter Optional (30)
--minlen INT (Trimmomatic) MINLEN parameter Optional (8)
--bowtie_index FILE (Bowtie) Path to index Optional ($baseDir/data/bowtie/human/human_index)
--bowtie_opts STR (Bowtie) additional options Optional (--sensitive -L 17)
--samtools_opts STR (samtools) options to view Optional (--no-PG -h -u -d 'NM' -e '![XS]')
--bowtie_threads INT Threads for bowtie Optional (7)
--fastqc_threads INT Threads for fastqc Optional (2)
--trimmo_threads INT Threads for trimmomatic Optional (3)
--samtools_threads INT Threads for samtools Optional (4)
--fastqcoutput FLAG Export fastqc logs (html and zip) Optional (false)
--trimoutput FLAG Export trimmomatic logs and trimmed fastq Optional (false)
--threeendcount FLAG Export 3'end read count Optional (false)
--bowtieoutput FLAG Export Bowtie's logs Optional (false)
--samtoolsoutput FLAG Export unique BAM files Optional (false)
--split FLAG Split count files by RNA Optional (false)
--scheduler STR Job scheduler Optional (slurm)
--qsize INT Max number of parallel jobs Optional (20)
--outdir DIR Output directory Optional (.)
--logdir DIR Log directory Optional ($outdir)
--help FLAG Displays this help
By default, the pipeline will only copy 5' read-end count files (See Read-end count files and two quality control reports in the output directory :
- multiqc_report.html : A RNA-seq report generated by MultiQC.
- rms_report.html : A RiboMethSeq report generated (See RiboMethSeq (RMS) Report)
Example output :
ribomethseq
├── counts
│ └── 5p
│ ├── sample1_R1_001.5_counts.csv
│ ├── sample2_R1_001.5_counts.csv
│ └── sample3_R1_001.5_counts.csv
├── multiqc_report.html
└── rms_report.html
Other directories can be copied by the pipeline in the output directory :
- bowtie2 : BAM alignment files and logs from Bowtie 2. (
--bowtieoutput) - counts/3p : 3' read-end count files. (
--threeendcount) - trimmomatic : Trimmed fastq files and logs. (
--trimoutput) - fastqc : fastqc report for each sample in both zip and html formats. (
--fastqcoutput)
The read-end count files represent the main output from this pipeline and are stored in the counts directory. By default, only the 5'end-read counts are exported (one file per sample). 3'end-read counts can also be exported alongside, using the --threeandcount parameter.
The files are in CSV format and have the following structure :
| Ref RNA | position on ref RNA | 5/3' read en count |
|---|---|---|
| NR_046235.3_5.8S | 1 | 735 |
| NR_046235.3_5.8S | 2 | 173 |
| NR_046235.3_5.8S | 3 | 59 |
| NR_046235.3_5.8S | 4 | 32 |
| NR_046235.3_5.8S | 5 | 21 |
The column headers have been added in the above example for clarity, but the real output files do not have them.
The RiboMethSeq QC report is stored in rms_report.html, at the root of the output directory.
It currently contains the following analyses :
- A end-read count boxplot and RLE for each samples.
- A distance heatmap to compare coverage profiles between samples.
- A correspondence analysis of the coverage profiles.