Template creation guidelines

Though none of these guidelines are mandatory nor required, their usage is highly recommended for several reasons:

  • Consistency in the outputs of the templates throughout the pipeline, particularly the status and report dotfiles (see Dotfiles section);
  • Debugging purposes;
  • Versioning;
  • Proper documentation of the template scripts.

Preface header

After the script shebang, a header with a brief description of the purpose and expected inputs and outputs should be provided. A complete example of such description can be viewed in flowcraft.templates.integrity_coverage.


Purpose section contains a brief description of the script’s objective. E.g.:


This module is intended parse the results of FastQC for paired end FastQ \

Expected input

Expected input section contains a description of the variables that are provided to the main function of the template script. These variables are defined in the input channels of the process in which the template is supposed to be executed. E.g.:

Expected input

The following variables are expected whether using NextFlow or the
:py:func:`main` executor.

- ``mash_output`` : String with the name of the mash screen output file.
    - e.g.: ``'sortedMashScreenResults_SampleA.txt'``

This means that the process that will execute this channel will have the input defined as:

file(mash_output) from <channel>

Generated output

Generated output section contains a description of the output files that the template script is intended to generated. E.g.:

Generated output

The generated output are output files that contain an object, usually a string.

- ``fastqc_health`` : Stores the health check for the current sample. If it
    passes all checks, it contains only the string 'pass'. Otherwise, contains
    the summary categories and their respective results

These can then be passed to the output channel(s) in the nextflow process:

file(fastqc_health) into <channel>


Since templates can be re-used by multiple processes, not all generated outputs need to be passed to output channels. Depending on the job of the nextflow process, it may catch none or all of the output files generated by the template.

Versioning and logging

FlowCraft has a specific logger (get_logger()) and versioning system that can be imported from flowcraft.templates.flowcraft_utils:

# the module that imports the logger and the decorator class for versioning
# of the script itself and other software used in the script
from flowcraft_utils.flowcraft_base import get_logger, MainWrapper


A logger function is also required to add logs to the script. The logs are written to the .command.log file in the work directory of each process.

First, the logger must be called, for example, after the imports as follows:

logger = get_logger(__file__)

Then, it may be used at will, using the default logging levels . E.g.:

logger.debug("Information tha may be important for debugging")
logger.info("Information related to the normal execution steps")
logger.warning("Events that may require the attention of the developer")
logger.error("Module exited unexpectedly with error:\\n{}".format(

MainWrapper decorator

This MainWrapper class decorator allows the program to fetch information on the script version, build and template name. For example:

# This can also be declared after the imports
__version__ = "1.0.0"
__build__ = "15012018"
__template__ = "process_abricate-nf"

The MainWrapper should decorate the main function of the script. E.g.:

def main():
    #some awesome code

Besides searching for the script’s version, build and template name this decorator will also search for a specific set of functions that start with the substring __get_version. For example:

def __get_version_fastqc():


    cli = ["fastqc", "--version"]
    p = subprocess.Popen(cli, stdout=PIPE, stderr=PIPE)
    stdout, _ = p.communicate()

    version = stdout.strip().split()[1][1:].decode("utf8")

    except Exception as e:
        version = "undefined"

    # Note that it returns a dictionary that will then be written to the .versions
    # dotfile
    return {
        "program": "FastQC",
        "version": version,
        # some programs may also contain build.

These functions are used to fetch the version, name and other relevant information from third-party software and the only requirement is that they return a dictionary with at least two key:value pairs:

  • program: String with the name of the program.
  • version: String with the version of the program.

For more information, refer to the build_versions() method.

Nextflow .command.sh

When these templates are used as a Nextflow template they are executed as a .command.sh file in the work directory of each process. In this case, we recommended the inclusion of an if statement to parse the arguments sent from nextflow to the python template. For example, imagine we have a path to a file name to pass as argument between nextflow and the required template:

# code check for nextflow execution
if __file__.endswith(".command.sh"):
    FILE_NAME = '$Nextflow_file_name'
    # logger output can also be included here, for example:
    logger.debug("Running {} with parameters:".format(
    logger.debug("FILE_NAME: {}".format(FILE_NAME))

Then, we could use this variable as the argument of a function, such as:

def main(FILE_NAME):
    #some awesome code

This way, we can use this function with nextflow arguments or without them, as is the case when the templates are used as standalone modules.

Use numpy docstrings

FlowCraft uses numpy docstrings to document code. Use this link for reference.