See here
See here
See here
Nextflow parameters can be provided in one of two ways:
- They can be specified in configuration files
- They can be specified on the command-line
- They can be specified in main.pbs
For example, all of the following are equivalent:
- Config file
params.reads = '/path/to/reads.csv'
params.readlength = 48
params.singleEnd = trueOR
params {
reads = '/path/to/reads.csv'
readlength = 48
singleEnd = true
}See configuration scopes for more information on this^
- Specifying parameters on the command-line
nextflow run main.nf --reads /path/to/reads.csv --readlength 48 --singleEnd trueParameters specified on the command-line (or in main.pbs) take precedence over those specified in configuration files. It is generally best-practice to have your parameters saved in a configuration file as this makes your analysis more reproducible if you need to run it again.
Profiles are configuration that can be included by specifying the profile name on the command-line. This CANNOT be set from the configuration file. For example, -profile sumner to include configuration specific to JAX's HPC Sumner.
Strings can be specified using 'single' or "double quotes"
File paths can be any one of the following:
- Local path - from directory that Nextflow is run in
- Full path
- Links -
https,ftp,s3&gslinks can all be used to specify input files provided that you have access to the file. Nextflow will automatically download these files into theworkdirectory (inwork/stage). On subsequent executions the pre-downloaded files will be used.
Integers can be specified without using quotes both in the configuration files and on the command-line
Boolean parameters can be set to either true or false. Many of the parameters are initialised to false in nextflow.config. You can set parameters to true on the command line just by using the flag. For example, just using --singleEnd will set the singleEnd parameter to true.
Side note:
However, be careful doing this as --singleEnd false will actually set the singleEnd parameter to the string 'false' not the boolean false. Counterintuively, as this is a string that is present it actually mean that singleEnd will evaluate to true 😆
This is another reason why it can be best to specify parameters in a confugration file rather than on the command-line
Memory units eg for max_memory can be specified in gigabytes eg 8.GB
Time units eg for max_time can be specified in hours eg 2.h
Both of these should be specified without quotes
Input files:
--reads Path to reads.csv file, which specifies the sample_id and path to FASTQ files
for each read or read pair (path).
This file is used if starting at beginning of pipeline. It can be file paths,
s3 links or ftp link.
(default: no reads.csv)
--bams Path to bams.csv file which specifies sample_id and path to BAM and BAM.bai
files (path)
If this file is provided, pipeline will start at Stringtie (and proceed through
rMATS and post processing).
(default: no bams.csv)
--rmats_pairs Path to rmats_pairs.txt file containing b1 (and b2) samples names (path)
(default: no rmats_pairs specified)
--run_name User specified name used as prefix for output files
(defaut: no prefix, only date and time)
--download_from Database to download FASTQ/BAMs from (available = 'TCGA', 'GTEX', 'SRA', 'FTP')
(string)
false should be used to run local files on the HPC (Sumner).
'TCGA' can also be used to download GDC data including HCMI data.
(default: false)
--key_file For downloading reads, use TCGA authentication token (TCGA) or
credentials.json file in case of 'GTEX'.
(default: false)
Main arguments:
--gtf Path to reference GTF file (path)
(default: no gtf specified)
--assembly_name Genome assembly name (available = 'GRCh38' or 'GRCm38', string)
(default: false)
--star_index Path to STAR index (path)
Star indices must be generated prior to run (with correct STAR version)
(default: false)
--singleEnd Specifies that the input is single-end reads (bool)
This parameter also automatically establishes the path to the SE or PE adapters.
For PE, set to false.
(default: false)
--stranded Specifies that the input is stranded ('first-strand', 'second-strand',
false (aka unstranded))
'first-strand' refers to RF/fr-firststrand in this pipeline.
(default: 'first-strand')
--readlength Read length - Note that all reads will be cropped to this length(int)
(default: no read length specified)
-profile Configuration profile to use. Can use multiple (comma separated, string)
On sumner, this should be set in the main.pbs or as a command-line parameter.
Profile can only be activated from the command line.
Available: base, docker, sumner, test and more.
Trimmomatic:
--minlen Drop the read if it is below a specified length (int)
Default parameters turn on --variable-readlength
To crop all reads and turn off --variable-readlength, set minlen = readlength
(default: 20)
--slidingwindow Perform a sliding window trimming approach (bool)
(default: true)
--adapter Path to adapter file (path)
(default: TruSeq3 for either PE or SE, see singleEnd parameter)
Star:
--mismatch Number of allowed mismatches per read (SE) or combined read (PE) (int)
SE ex. read length of 50, allow 2 mismatches per 50 bp
PE ex. read length of 50, allow 2 mismatches per 100 bp
(default: 2)
--overhang Overhang (int)
(default: readlength - 1)
--filterScore Controls --outFilterScoreMinOverLread and outFilterMatchNminOverLread
For TCGA values:
https://docs.gdc.cancer.gov/Data/Bioinformatics_Pipelines/Expression_mRNA_Pipeline/
(default: 0.66)
--sjdbOverhangMin Controls --alignSJDBoverhangMin (int)
For TCGA values:
https://docs.gdc.cancer.gov/Data/Bioinformatics_Pipelines/Expression_mRNA_Pipeline/
(default: 3)
--soft_clipping Enables soft clipping (bool)
If true, the STAR parameter will be --alignEndsType 'Local' and the rMATS parameter
--allow-clipping will be added.
If false, the STAR parameter will be --alignEndsType 'EndToEnd' and no rMATS
parameter is added.
NOTE: Soft Clipping will cause read lengths to be variable, so turn soft_clipping
off if reads need to be same length. Variable read length parameter is turned on
in rMATS when minlen does not equal readlength.
(default: true)
--save_unmapped Save unmapped and partially mapped reads in separate file (bool)
(default: false)
--star_memory Max memory to be used by STAR to sort BAM files.
(default: Available task memory)
rMATS:
--statoff Skip the statistical analysis (bool)
If using only b1 as input, this must be turned on.
(default: false)
--paired_stats Use the paired stats model (bool)
(default: false)
--novelSS Enable detection of unnanotated splice sites (bool)
(default: false)
--mil Minimum Intron Length. Only impacts --novelSS behavior (int)
(default: 50)
--mel Maximum Exon Length. Only impacts --novelSS behavior (int)
(default: 500)
Other:
--test For running trim test (bool)
To run the first half of the pipeline (through STAR), set test = true.
(default: false)
--max_cpus Maximum number of CPUs (int)
(default: 72)
--max_memory Maximum memory (memory unit)
(default: 760.GB)
--max_time Maximum time (time unit)
(default: 72.h)
--skiprMATS Skip rMATS (bool)
(default: false)
--skipMultiQC Skip MultiQC (bool)
(default: false)
--outdir The output directory where the results will be saved (string)
On Sumner, this must be set in the main.pbs or via command line.
NF_splicing_pipeline.config will not overwrite main.pbs.
(default: <directory where you submit the job>/results)
--mega_time Sets time limit for processes withLabel 'mega_memory' in the main.nf using the
base.config (time unit)
Make sure '#SBATCH -t' in 'main.pbs' is appropriately set if you are changing this parameter.
(default: 20.h)
--gc_disk_size Only specific to google-cloud executor. Adds disk-space for few aggregative processes.
(default: "200 GB" based on 100 samples. Simply add 2 x Number of Samples)
--debug This option will enable echo of script execution into STDOUT with some additional
resource information (such as machine type, memory, cpu and disk space)
(default: false)
--error_strategy Mode of pipeline handling failed processes.
Possible values: 'terminate', 'finish', 'ignore', 'retry'.
Check nextflow documentation for detailed descriptions of each mode:
https://www.nextflow.io/docs/latest/process.html#process-page-error-strategy
Set this parameter in the main.pbs, on the command line, or see NF_splicing_pipeline.config
example (does not work like normal config param)
This does not overwrite CloudOS config, which is set to:
'errorStrategy = { task.exitStatus in [3,9,10,14,143,137,104,134,139] ? 'retry': 'ignore'}
(default (non-cloudos): 'finish')
--cleanup This option will enable nextflow work folder cleanup upon pipeline successfull
completion. All intermediate files from nexftlow processes' workdirs will be
cleared, staging folder with staged files will not be cleared.
If pipeline is completed with errors or interrupted cleanup will not be executed.
Following successfull run resumed from the failed run with --cleanup option enabled
will only clear folders of processess created in the latest run, it will not clear
cached folders coming from previous pipleine runs.
Set this parameter in the main.pbs, on the command line, or see NF_splicing_pipeline.config
example (does not work like normal config param)
(default non-cloudos: true; cloudos: false)
Whereas parameters are set on the command-line using double dash options eg --reads, parameters passed to Nextflow itself can be provided with single-dash options eg -profile.
You can see some of these options here in the Nextflow documentation.
Some useful ones include (specified in main.pbs):
-resumewhich will resume any cached processes that have not been changed-with-traceeg-with-trace trace.txtwhich gives a trace report for resource consumption by the pipeline-with-dageg-with-dag flowchart.pngwhich produces the DAG visualisation graph showing each of the different processes and the connections between them (the channels)