def __init__(self, job_name, out_sh, directory, submit,
samtools_sort_args=''):
# samtools_sort_args = cl.args['samtools_sort_args']
# directory = cl.args['dir']
# add final forward slash if it's not the current directory and it's
# not the empty string. May cause bugs if your current directory is
# the base directory '/', but I'm not too worried about that :)
if directory != '' and not directory.endswith('/'):
directory += '/'
for sam in glob('%s*.sam' % directory):
bam = sam.replace('.sam', '.bam')
qsub_commands = []
# Samtools view flags:
# '-b': output is BAM
# '-S': input is SAM
qsub_commands.append('samtools view -bS %s > %s' % (sam, bam))
sorted_prefix = bam.replace('.bam', '.sorted')
qsub_commands.append('samtools sort %s %s %s' %
(samtools_sort_args, bam, sorted_prefix))
# submitter_prefix = 'sam2bam_sort_index_%s' % (bam)
# submitter_sh = submitter_prefix + '.sh'
sub = qtools.Submitter(queue_type='PBS',
sh_filename=out_sh,
commands=qsub_commands,
job_name=job_name, nodes=1, ppn=16,
queue='home')
sub.write_sh(submit=submit)
def __init__(self,
gtf,
gff,
index_dir,
job_name,
out_sh,
submit,
index_gff_py='index_gff.py'):
gtf_or_gff = gtf if gtf is not None else gff
# qs.add_Q_resource('-l', 'bigmem')
# qs.add_Q_resource('-l', 'h_vmem=30G')
# qs.add_Q_resource('-e', submitter_err)
# qs.add_Q_resource('-o', submitter_out)
commands = []
if gtf is not None:
gff = gtf.replace('gtf', 'gff')
gtf2gff = 'gtf2gff3.pl %s > %s' % (gtf, gff)
commands.append(gtf2gff)
command = '%s --index %s %s' % (index_gff_py, gff, index_dir)
commands.append(command)
qs = qtools.Submitter(queue_type='PBS',
sh_filename=out_sh,
commands=commands,
job_name=job_name,
nodes=1,
ppn=1,
queue='home',
walltime='0:30:00')
qs.write_sh(submit=submit)
def __init__(self,
sample_info_file,
submit=True,
out_sh=None,
queue_type=None):
sInfo = pd.read_table(sample_info_file)
cmds = []
for row, dat in sInfo.iterrows():
id = dat['Sample ID']
bam = dat['Bam File']
try:
species = dat['Species']
except:
species = "hg19"
try:
strand = dat['Strand']
assert strand in ['flip', 'sense',
'both'] #antisense, sense, run both
except:
strand = 'both'
out = id + ".splices"
if (strand == 'sense') or (strand == 'both'):
oldsplice_command = "oldsplice.py -b %s -s %s -o %s --splice_type SE --splice_type MXE --processors 16" % (
bam, species, out)
cmds.append(oldsplice_command)
if (strand == 'flip') or (strand == 'both'):
oldsplice_command = "oldsplice.py -f -b %s -s %s -o %s --splice_type SE --splice_type MXE --processors 16" % (
bam, species, out.replace(".splices", ".flip.splices"))
cmds.append(oldsplice_command)
sh_filename = 'runOldsplice.sh' if out_sh is None else out_sh
sub = qtools.Submitter(array=True,
queue="home",
nodes=1,
commands=cmds,
sh_filename=sh_filename,
job_name="oldsplice",
max_running=1000,
ppn=16,
queue_type=queue_type)
sub.job(submit=submit)
from gscripts import qtools
import pandas as pd
import sys
sInfo = pd.read_table(sys.argv[1])
cmds = []
Sub = qtools.Submitter()
for row, dat in sInfo.iterrows():
id = dat['Sample ID']
bam = dat['Bam File']
try:
species = dat['Species']
if species == "hg19":
if "gff" in dat:
gff = dat['gff']
else:
gff = "/home/yeo-lab/genomes/hg19/miso_annotations/SE.hg19.gff3"
elif species == "panTro4":
if 'gff' in dat:
gff = dat['gff']
else:
gff = '/oasis/tscc/scratch/mlovci/apeSeq/genomes/all_genomes/annotations/chimp/events/commonshortest/SE.panTro4.gff3'
'
else:
raise ValueError("I don't know where species %s's gff file is" % (dat['Species']))