Files
galaxy/scripts/microbes/get_builds_lengths.py
T
Nicola Soranzo 9d74bba7fb Drop support for retired Python 3.5
Upgrade syntax using `pyupgrade --py36-plus` .

Manually drop several `six` imports.

Also:
- Remove broken pr_cache in scripts/bootstrap_history.py
- Fix broken prefix removal in lib/galaxy/tool_util/deps/mulled/mulled_build.py
2020-10-07 11:52:13 +01:00

61 lines
2.1 KiB
Python

#!/usr/bin/env python
# Dan Blankenberg
import os
import sys
assert sys.version_info[:2] >= (2, 6)
def __main__():
base_dir = os.path.join(os.getcwd(), "bacteria")
try:
base_dir = sys.argv[1]
except IndexError:
pass
organisms = {}
for result in os.walk(base_dir):
this_base_dir, sub_dirs, files = result
for file in files:
if file[-5:] == ".info":
tmp_dict = {}
info_file = open(os.path.join(this_base_dir, file))
info = info_file.readlines()
info_file.close()
for line in info:
fields = line.replace("\n", "").split("=")
tmp_dict[fields[0]] = "=".join(fields[1:])
if 'genome project id' in tmp_dict.keys():
name = tmp_dict['genome project id']
if 'build' in tmp_dict.keys():
name = tmp_dict['build']
if name not in organisms.keys():
organisms[name] = {'chrs': {}, 'base_dir': this_base_dir}
for key in tmp_dict.keys():
organisms[name][key] = tmp_dict[key]
else:
if tmp_dict['organism'] not in organisms.keys():
organisms[tmp_dict['organism']] = {'chrs': {}, 'base_dir': this_base_dir}
organisms[tmp_dict['organism']]['chrs'][tmp_dict['chromosome']] = tmp_dict
for org in organisms:
org = organisms[org]
# if no gpi, then must be a ncbi chr which corresponds to a UCSC org, w/o matching UCSC designation
try:
build = org['genome project id']
except KeyError:
continue
if 'build' in org:
build = org['build']
chrs = []
for chrom in org['chrs']:
chrom = org['chrs'][chrom]
chrs.append("{}={}".format(chrom['chromosome'], chrom['length']))
print("{}\t{}\t{}".format(build, org['name'], ",".join(chrs)))
if __name__ == "__main__":
__main__()