- Input file naming convention
- Launching an analysis
- Resulting folder
- Writing a whole configuration file to create a pipeline
- Providing an order
- Sending software parameters
- Cleaning steps
- Compressing steps
- Scheduler and parallel runs
- Providing a specific environment variables
- Transferring files by scp
- Generic command
- Output and error logs
- Drawing the pipeline
- Report Option
- Process_radstacks steps
The TOGGLe v0.3.x version allows users to create their own customized pipelines. You can modify not only the different options for each software but also define dedicated pipelines for your analyses.
You can therefore switch, add and remove steps, start from FASTA/FASTA.GZ, FASTQ/FASTQ.GZ, SAM/BAM, BED, GFF or VCF/VCF.GZ files, ask for individual treatments only, individual then common (such as individual mapping followed by common calling), or even only common treatments.
TOGGLE will automatically assign sample name (readgroup) based on the input file name structure, picking up the “text” before the first dot.
- individual1_1.fastq will be understood as individual1
- mapping1.sam will be understood as mapping1
- myVcf.complete.vcf will be understood as myVcf
For Fastq files, only one underscore (‘_’) is allowed, before the direction of sequences ( _1, _2 or _R1, _R2), just before the extension (.fastq).
Please use only UTF-8 standard symbols, no weird characters or space, pipe, tilde or any type of commas.
The current version is based on the toggleGenerator.pl script
Thus for instance, a classical command would be:
All the the paths (files and folders) can be provided as absolute (/home/mylogin/data/myRef.fasta) or relative (../data/myRef.fasta or ~/data/myRef.fasta).
|Required named arguments:|
|-d / –directory DIR:||a folder with raw data to be treated (FASTQ, FASTQ.GZ, SAM, BAM, VCF, VCF.GZ, BED, GFF3, GTF, TXT)|
|-c / –config FILE:||generally it is the software.config.txt file but it can be any text file structured as shown below.|
|-o / –outputdir DIR:||the folder were results will be outputted. Must be empty. If it does not exist, TOGGLe will create it.|
|Optional named arguments:|
|-r / –reference FILE:||the reference FASTA file to be used. (1)|
|-g / -gff FILE:||the GFF/GTF file to be used for some tools.|
|-k / –keyfile FILE:||the keyfile use for demultiplexing step.|
|-add / –add:||use if you want to add new samples to an already run analysis.|
|-rerun / –rerun:||use if you want to re-run samples that have encountered error previously.|
|-nocheck / –nocheckFastq:||by default TOGGLe checks if input format is correct in every file. This option allows to skip this step.|
|-report / –report:||generate pdf report (more info)|
|-v / –version:||Use if you want to know which version of TOGGLe you are using|
|-h / –help:||show help message and exit|
(1): If no reference indexes exist, they will be created accordingly to the pipeline requested indexes. If they exist, they will not be re-created UNLESS the pipeline order (see below) expressively requests it (to update the index e.g.)
TOGGLe will generate an output folder containing different files and subfolders, as follows:
The final results are contained in the finalResults folder. TOGGLe will also copy the software config (and reference if any) files corresponding to the analysis, in order users can recover their options. The output folder contains all sub analyses, i.e. the individual analyses or intermediate data.
The toggleGenerator.pl script will use the configuration file informations (here named as software.config.txt file) to create specific pipeline scripts in the output folder, called toggleBzz.pl for individual treatments (individual mapping e.g.) and toggleMultiple.pl for global treatment (global SNP calling e.g.).
The order will be checked before creating the scripts, to control that the output files from the step n-1 will be accepted as input for the step n.
The order of a specific pipeline is provided in the software.config.txt as the software order
Thus, if you provide options such as:
You will obtain a pipeline taking FASTQ files as input (plain text of gzipped), aligning them upon a given reference using bwa aln/sampe, and finally cleaning the resulting sam and providing a BAM file (if samtools view option are as such).
Note that the way you write the name of the step is not really important, e.g. it is not case-sensitive, and can have space between the different words. A dedicated module will adjust everything.
You can start from any type of format recognized by TOGGLE, and from any step. Thus, if you have a raw VCF file you can start from this one at step one and clean it as follows (even if your VCF contains informations for multiple individuals):
BE CAREFUL: You can comment any step using a ‘#’ symbol before it (ex #3=bwaAln) to avoid to retype all numbers. However, such a comment will provoke errors in cleaning and compressing steps. The best approach is to provide a list of following numbers (e.g. 1, 2, 3, 4 and not 1, 3, 4). This is also true for steps higher than 1000.
If you want to adress multiple times the same step BUT with different configurations (e.g. a first samtools view to obtain a BAM then a second samtools view to clean this BAM using the -f option), you have to significate it in the order key as follows:
In the same time you have to provide the same informations in your configuration:
BE CAREFUL: in such multiple times repeated software, you have to put a space between the software name and the step (ex samtoolsview 1 and samtoolsView 2) in the order as well as in the options !!
If you want for instance to map all individuals (multiple files) separately then perform a common SNP calling, please add a step number higher than 999:
This pipeline will map FASTQ then sort per coordinates the SAM for each individuals, and then will launch a common step for all as a global SNP calling. You can add after this calling other steps (1001=gatkVariantFiltrator for example).
Warnings: if you want to map and call SNP from only one sample, it is not necessary to call a 1000+ step (here gatkUnifiedGenotyper would be then the 4th step).
If you want only a global treatment for multiple files (you may have all your formatted BAM and you want to perform the calling and subsequent steps), you can create the following pipeline:
The pipeline will treat the data as a global pipeline in this case, without separating the individual files.
Any software configuration will start as follows:
You can address any option to any given software (as soon as the given option exists for the given software ^^ ) as follows:
You can also write as:
The software name is not case sensitive and the subprogram can be “glued” to the name (bwaALN is recognized, as well as bwa aln).
If your option has an equal symbol within, such as -l hostname=MyNode, you have to write the option as follows:
BETA FEATURE: Using the cleaning steps may impair the execution of some steps. Use it carefully.
In order to gain hard drive space, you may want to remove some steps from your complete analysis.
In this case, you can specify in the configuration file which step must be ERASED along the pipeline (as soon as the step following them has been correctly completed), using the key cleaner.
As an example
There only the last step result (samtools sort) will be conserved. The folders and logs of the cleaned steps are conserved, but not the resulting files.
NOTE: the last step will never be cleaned. This is true for single analyses (steps lower than 999 - toggleBzz.pl script) as well as for common steps (higher than 1000 - toggleMultiple.pl script)
BETA FEATURE: Using the compressing steps may impair the execution of some steps. Use it carefully.
In order to gain hard drive space but conserving data, you may want to compress some steps from your complete analysis.
In this case, you can specify in the configuration file which step must be COMPRESSED in tar.gz along the pipeline (as soon as the step following them has been correctly completed), using the key compress.
As an example
There only the last step result (samtools sort) will be conserved, the other being compressed in their respective tar.gz archive.
NOTE: as for the cleaner system, the last step will never be compressed. This is true for single analyses (steps lower than 999 - toggleBzz.pl script) as well as for common steps (higher than 1000 - toggleMultiple.pl script).
BE CAREFUL: CLEANING IS OF HIGHER ORDER THAN COMPRESS. If the same step is required to be cleaned AND compressed, it will be only cleaned!
TOGGLe is scheduler-aware (SGE, MPRUN, Slurm and LSF on v0.3+), and is able to decide by itself to launch on such a system. However, the jobs will be launched in parallel only if the software.config file contains informations for scheduling, i.e. the queue name, number of core/threads per jobs, etc…
For SGE, configuration can be provided such as:
If the software.config file contains those SGE/Slurm/MPRUN/LSF informations AND the machine is SGE/Slurm/MPRUN/LSF capable, the toggleBzz.pl and the toggleMultiple.pl scripts will be launched as parallel jobs.
Moreover, in parallel as in linear mode, an error in one individual will not stop the whole process. This individual will be marked as error in the error log, but the others will continue.
If you need to provide specific ENVIRONMENT variables in your jobs (e.g. export or module load), you can provide the $env key in your software.config file
If you need to transfer the files to a temporary disk (e.g. /scratch) using scp, you can provide a $scp key followed by the temporary directory name in your configuration file
BETA FEATURE: if you need to launch a tool not yet integrated in TOGGLe, you can provide a Generic command using the following syntax (example of simple bash command to count SNP in a VCF file)
The FILEIN and FILEOUT terms must be written such as, and will be automatically replaced by TOGGLe. Here in the example, the FILEIN will be modified as the previous VCF file, and FILEOUT as the output file.
BE CAREFUL: NO FORMAT CHECK will be performed, as well as no sanity control for the following step. This option is not recommanded to non-expert users.
TOGGLe will generate two main types of logs, the .o for normal output and the .e for the errors and warnings (these last ones are normally empty files). Each level of TOGGLe will generate this pair of log:
TOGGLE_date_log.o/.e logs represent the general output for the complete analysis (toggleGenerator.pl logs). They are located at the root of the output directory.
SampleName_log.o/.e logs represent the local output for sub analysis (toggleBzz.pl and toggleMultiple.pl logs). They are located in their respective subdirectories in the output folder.
If Graphviz is installed on the running server, the toggleGenerator.pl script will also generate a graphical view of the pipeline in the output folder. If Graphviz is not installed, a .dot file is nevertheless created, allowing the user to create a graphical view on another machine having Graphviz.
If the -report option is set, TOGGLe will perform a report in PDF providing all the informations about the execution of the pipeline:
- Command line provided
- Software version
- Basic stats about input files
- Basic stats about the mapping, VCF, and BAM files
If you want to use this option, the system must have access to GraphViz and LaTEX distribution (TexLive recommanded)
If process_radstacks from stacks is used in your pipeline (for demultiplexing), you have to put it as the first step or TOGGLe will return an error and stop. After demultiplexing, TOGGLe will start from step 2 to treat all demultiplexed individuals.