def runAllIndelMap(start_idx=0,
stop_idx=10000000,
overbeek_only=False,
queue='normal',
map_dir='/mapped_reads/',
max_cut_dist=4,
num_parts=1,
order_by_incomplete=False):
all_dir, out_dir = getAllDataDirs(), getLogDir()
if overbeek_only: print('Computing for Overbeek guides only')
completed_lookup = {}
if order_by_incomplete:
f = io.open('../quality_checks/status.log')
completed_lookup = {
toks[0]: min([eval(x) for x in toks[1:]]) != 0
for toks in csv.reader(f, delimiter='\t')
}
f.close()
completed = [
x for x in all_dir if getDirLabel(x) in completed_lookup
and completed_lookup[getDirLabel(x)]
]
not_completed = [x for x in all_dir if x not in completed]
max_files_per_dir = 20
file_per_part = int(max_files_per_dir / num_parts + 0.99)
i, idx = 0, 0
for dirname in not_completed + completed:
if len(not_completed) == i:
print('-------------------------------------------------')
i += 1
print(getShortDir(dirname), idx)
if not os.path.isdir(dirname + map_dir): continue
for subdir in getSubdirs(dirname, withpath=False):
if overbeek_only and subdir != 'Oligos_71': continue
args = (dirname, getNullDir(dirname), subdir, file_per_part,
map_dir, max_cut_dist, getIndelMapExe(), getPythonCmd())
cmd = getPythonCmd(
) + ' indelmap_subdir.py %s %s %s %d - 0 %s %d %s %s' % args
idx = runCmdCheckIdx(cmd,
idx,
start_idx,
stop_idx,
out_dir,
'out_indelmap_%s' % getDirLabel(dirname),
numj=num_parts,
queue=queue)
def runPerSubdir(python_script,
out_label,
caller,
extra_args='',
include_null=False):
idx = 0
for dirname in [
x for x in getAllDataDirs() if include_null or not isNullDir(x)
]:
if not os.path.isdir(dirname + '/mapped_reads'):
print(getShortDir(dirname)), 'No mapped_reads directory'
else:
subdirs = getSubdirs(dirname)
idx = runSubdir(idx,
subdirs,
getShortDir(dirname),
python_script,
out_label,
caller,
extra_args=extra_args)
from selftarget.oligo import getOligoIdsFromFile
from selftarget.data import getDirLabel, getIndelSummaryFiles, getSubdirs
def filterLargeI(profile):
return {x:profile[x] for x in profile if (x[0] == '-' or x[0] != 'I' or x[1] != '1')}
if len(sys.argv) != 4:
print('compare_pairwise.py <dirname1> <dirname2> <subdir>')
else:
remove_largeI = True
dirname1, dirname2 = sys.argv[1], sys.argv[2]
subdir = sys.argv[3]
if subdir not in getSubdirs(dirname1, withpath=False):
raise Exception('No subdir %s in %s' % (subdir, dirname1) )
if subdir not in getSubdirs(dirname2, withpath=False):
raise Exception('No subdir %s in %s' % (subdir, dirname2) )
out_dir = 'profile_comparison_summaries' if not remove_largeI else 'profile_comparison_summaries_nolargeI'
if not os.path.isdir(out_dir): os.mkdir(out_dir)
out_dir += '/%s_vs_%s' % (getDirLabel(dirname1),getDirLabel(dirname2))
if not os.path.isdir(out_dir): os.mkdir(out_dir)
out_file = out_dir + '/%s.txt' % subdir
fout = io.open(out_file,'w')
fout.write(u'ID\tNum Reads 1\tNum Reads 2\tNum States 1\tNum States 2\tNum null reads 1\tNum null reads 2\tKL with Null\tKL without null\tPerc Accepted Reads 1\tPerc Accepted Reads 2\t1st Nonmatch Indel\tNum Top 3 Common\tNum Top 5 Common\tNum Top 10 Common\tProfile 1 Entropy (before mods)\tProfile 2 Entropy (before mods)\tProfile 1 Entropy (after mods)\tProfile 2 Entropy (after mods)\tPerc Overlap\tP1 Perc in Top 3\tP2 Perc in Top 3\tP1 Perc in Top 5\tP2 Perc in Top 5\tP1 Perc in Top 10\tP2 Perc in Top 10\n')
dir1_files = getIndelSummaryFiles(dirname1 + '/mapped_reads/' + subdir, withpath=False)
dir2_files = getIndelSummaryFiles(dirname2 + '/mapped_reads/' + subdir, withpath=False)
from selftarget.util import getLogDir, runCmdCheckIdx, runSubdir
if __name__ == '__main__':
all_dir, out_dir = getAllDataDirs(), getLogDir()
idx = 0
for old_lib in [False]: #[True,False]:
lib_dirs = [
x for x in all_dir
if (isOldLib(x) == old_lib) and os.path.isdir(x + '/mapped_reads')
and 'DPI7' in x and 'K562_1600x_LV7B_DPI7' not in x
and '2A_TREX' not in x and 'K562_800x_7A_DPI7_may' not in x
]
for dirname1, dirname2 in itertools.combinations(lib_dirs, 2):
subdirs_1 = getSubdirs(dirname1, withpath=False)
subdirs_2 = getSubdirs(dirname2, withpath=False)
common_subdirs = set(subdirs_1).intersection(set(subdirs_2))
label = '%s\t%s' % (getShortDir(dirname1), getShortDir(dirname2))
extra_args = '%s %s ' % (dirname1, dirname2)
idx = runSubdir(idx,
common_subdirs,
label,
'compare_pairwise.py',
'out_compare_pairwise',
__file__,
extra_args=extra_args)
if len(sys.argv) >= 4: overbeek_only = eval(sys.argv[3])
if len(sys.argv) >= 5: queue = sys.argv[4]
all_dir, out_dir = getAllDataDirs(), getLogDir()
if overbeek_only: print 'Computing for Overbeek guides only'
file_per_part = 1
max_files_per_dir = 20
num_parts = (max_files_per_dir + file_per_part - 1) / file_per_part
i, idx = 0, 0
for dirname in all_dir:
i += 1
print getShortDir(dirname), idx
if not os.path.isdir(dirname + '/mapped_reads'): continue
for subdir in getSubdirs(dirname, withpath=False):
if overbeek_only and subdir != 'Oligos_71': continue
cmd = './run_correct_indel.sh %s %s %s %d' % (
dirname, getNullDir(dirname), subdir, file_per_part)
idx = runCmdCheckIdx(cmd,
idx,
start_idx,
stop_idx,
out_dir,
'out_correct_indelmap_%s' % getDirLabel(dirname),
numj=num_parts,
queue=queue)