SConstruct-get-data

import os
from os import path
import sys

import SCons
from SCons.Script import (Environment, Variables, Help, Decider, Dir)
from bioscons.fileutils import rename

import common

# TODO: move to bioscons
SCons.Script.EnsureSConsVersion(3, 0, 1)

# Ensure that a virtualenv is active before importing non-stdlib dependencies.
if not os.environ.get('VIRTUAL_ENV'):
    sys.exit('--> an active virtualenv is required')

# check timestamps before calculating md5 checksums
Decider('MD5-timestamp')

# ############### start inputs ################
user_args, conf = common.get_conf('settings.conf')

outdir = conf['get_data'].get('outdir', 'dada2')
plates = conf['get_data'].get('plates', '').strip().split()
assert bool(plates), 'must provide a value for conf["get_data"]["plates"]'
projects = conf['get_data'].get('projects', '').strip().splitlines()

refpkg = conf['get_data'].get('refpkg')

sample_info = [path.join('/fh/fast/fredricks_d/bvdiversity/data/dada2_nf_out',
                         plate, 'output', 'sample_information.csv')
               for plate in plates]

singularity = conf['singularity'].get('singularity', 'singularity')
dada2_img = conf['singularity']['dada2']
binds = [os.path.abspath(pth)
         for pth in conf['singularity']['binds'].strip().splitlines()]

# ############### end inputs ##################

vars = Variables()
vars.Add('out', '', outdir)
env = Environment(
    ENV=os.environ,
    variables=vars,
    SHELL='bash',
    cwd=os.getcwd(),
    singularity=singularity,
    binds=' '.join('-B {}'.format(pth) for pth in ['$cwd'] + binds),
    dada2_img=('$singularity exec $binds --pwd $cwd {}'.format(dada2_img)),
)

Help(vars.GenerateHelpText(env))

if projects:
    env['projects'] = '--projects ' + ' '.join(['"{}"'.format(p) for p in projects])

sample_info, seqtabs = env.Command(
    target=['$out/sample_info.csv', '$out/seqtabs.txt'],
    source=sample_info,
    action=('bin/gather_seqtabs.py $SOURCES '
            '--sample-info ${TARGETS[0]} '
            '--seqtabs ${TARGETS[1]} '
            '$projects')
)

sv_fa, sv_table, weight, specimen_map, sv_table_long, specimen_table = env.Command(
    target=['$out/seqs.fasta',
            '$out/dada2_sv_table.csv',
            '$out/weights.csv',
            '$out/specimen_map.csv',
            '$out/dada2_sv_table_long.csv',
            '$out/specimen_table.csv'],
    source=seqtabs,
    action=('/fh/fast/fredricks_d/bvdiversity/data/dada2-nf/bin/write_seqs.py '
            '--seqtablist $SOURCE '
            '--seqs ${TARGETS[0]} '
            '--sv-table ${TARGETS[1]} '
            '--weights ${TARGETS[2]} '
            '--specimen-map ${TARGETS[3]} '
            '--sv-table-long ${TARGETS[4]} '
            '--specimen-table ${TARGETS[5]} '
            '--direction merged '
            '-j 20')
)

if refpkg:
    local_refpkg = env.Command(
        target=Dir(rename(refpkg, pth='.')),
        source=refpkg,
        action='cp -r $SOURCE $TARGET'
    )