__author__ = "Johannes Köster, Felix Mölder, Christopher Schröder" __copyright__ = "Copyright 2017, Johannes Köster" __email__ = "[email protected], [email protected]" __license__ = "MIT" from os import path from snakemake.shell import shell from tempfile import TemporaryDirectory shell.executable("bash") log = snakemake.log_fmt_shell(stdout=False, stderr=True) params = snakemake.params.get("extra", "") norm = snakemake.params.get("normalize", False) # Infer output format uncompressed_bcf = snakemake.params.get("uncompressed_bcf", False) out_name, out_ext = path.splitext(snakemake.output[0]) if out_ext == ".vcf": out_format = "v" elif out_ext == ".bcf": if uncompressed_bcf: out_format = "u" else: out_format = "b" elif out_ext == ".gz": out_name, out_ext = path.splitext(out_name) if out_ext == ".vcf": out_format = "z" else:
# f = open(snakemake.log.run, 'at')
# f.write("## COMMAND: "+command+"\n")
# f.close()
# shell(command)
# command = "rm "+snakemake.input.bam + ".bai"
# f = open(snakemake.log.run, 'at')
# f.write("## COMMAND: "+command+"\n")
# f.close()
# shell(command)
# else:
TOOL = "umi_tools"
SAMTOOLS = "samtools"
shell.executable("/bin/bash")
version = str(
subprocess.Popen(SAMTOOLS + " --version 2>&1 | grep \"samtools\" ",
shell=True,
stdout=subprocess.PIPE).communicate()[0], 'utf-8')
f = open(snakemake.log.run, 'at')
f.write("## VERSION: " + version + "\n")
f.close()
command = TOOL + " dedup -I " + snakemake.input.bam + " -S " + snakemake.output.bam + " --log " + snakemake.output.mtx \
+ " --extract-umi-method=read_id --umi-separator='_' --method=directional --edit-distance-threshold=0 \
--spliced-is-unique --multimapping-detection-method=NH"
f = open(snakemake.log.run, 'at')
f.write("## COMMAND: " + command + "\n")
def init_powersnake():
# pipefail is supported by bash only.
shell.executable(os.popen('which bash').read().strip())
shell.prefix('set -e; set -o pipefail; ')
import os
from snakemake.shell import shell
from snakemake.utils import makedirs
shell.executable('/bin/bash')
URL = 'https://github.com/lcdb/lcdb-test-data/blob/add-chipseq/data/{}?raw=true'
def _download_file(fn, dest=None):
url = URL.format(fn)
if dest is None:
dest = fn
makedirs(os.path.dirname(dest))
basename = os.path.basename(fn)
shell('wget -q -O- {url} > {dest}')
return dest
_download_file('rnaseq_samples/sample1/sample1.small_R1.fastq.gz')
_download_file('rnaseq_samples/sample2/sample2.small_R1.fastq.gz')
_download_file('rnaseq_samples/sample3/sample3.small_R1.fastq.gz')
_download_file('rnaseq_samples/sample4/sample4.small_R1.fastq.gz')
shell('mkdir -p data/rnaseq_samples/sample{{1,2,3,4}}')
for n in [1, 2, 3, 4]:
shell(
'mv rnaseq_samples/sample{n}/sample{n}.small_R1.fastq.gz '
'data/rnaseq_samples/sample{n}/sample{n}_R1.fastq.gz'
)
shell('rm -r rnaseq_samples')
def init_powersnake():
# pipefail is supported by bash only.
shell.executable(os.popen('which bash').read().strip())
shell.prefix('set -e; set -o pipefail; ')
