Naming of reused GWF templates using groups

reused_gwf_templates_with_groups.py

from gwf import Workflow, AnonymousTarget
from groups import Group # separate Gist

def mask_template(path):
    inputs = {'path': path}
    outputs = {'path': path + '.masked'}
    options = {}
    spec = """./some_command {}""".format(path)
    return AnonymousTarget(inputs=inputs, outputs=outputs, options=options, spec=spec)

def step_one_template(path):
    inputs = {'path': path}
    outputs = {'path': path + '.first'}
    options = {}
    spec = """./some_command {}""".format(path)
    return AnonymousTarget(inputs=inputs, outputs=outputs, options=options, spec=spec)

def step_two_template(path):
    inputs = {'path': path}
    outputs = {'path': path + '.second'}
    options = {}
    spec = """./some_command {}""".format(path)
    return AnonymousTarget(inputs=inputs, outputs=outputs, options=options, spec=spec)

input_files = {'CEU': ['ceu_chr1.txt', 'ceu_chr2.txt'],
               'CHB': ['chb_chr1.txt', 'chb_chr2.txt'],
}

gwf = Workflow()    

with Group(gwf, prefix='with_repeats') as g: # <- NB g
    for pop, paths in input_files.items():
        with Group(g, prefix=pop) as gg:     # <- NB gg

            # så kan du lave dine maps som om der kun var en population:
            step_one_tasks = gg.map(step_one_template, paths, name='step_one')  # <- NB gg not gwf
            step_two_tasks = gg.map(step_two_template, step_one_tasks.outputs, name='step_two')

with Group(gwf, prefix='no_repeats') as g:
    for pop, paths in input_files.items():
        with Group(g, prefix=pop) as gg:

            # så kan du lave dine maps som om der kun var en population:
            mask_tasks = gg.map(mask_template, paths, name='mask')
            step_one_tasks = gg.map(step_one_template, mask_tasks.outputs, name='step_one')
            step_two_tasks = gg.map(step_two_template, step_one_tasks.outputs, name='step_two')