def get_gmt_url(species, gse, gpl):
"""
:type species: str
:type gpl: str
:type gse: str
"""
path_to_file = here('gmt', '{}_{}.gmt'.format(gse, gpl))
if getattr(settings, 'MEDIA_ARE_LOCAL', False):
path = here(settings.MEDIA_ROOT, path_to_file)
return here(settings.MEDIA_URL,
path_to_file) if os.path.exists(path) else None
return here(settings.MEDIA_URL, path_to_file)
def get_module_heat_map_url(species, gse, gpl, module_number):
"""
:type module_number: int
:type gpl: str
:type gse: str
:type species: str
:rtype: str
"""
path_to_image = here('heatmaps', species,
'{}_{}_module_{}.svg'.format(gse, gpl, module_number))
if getattr(settings, 'MEDIA_ARE_LOCAL', False):
path = here(settings.MEDIA_ROOT, path_to_image)
return here(settings.MEDIA_URL,
path_to_image) if os.path.exists(path) else None
return here(settings.MEDIA_URL, path_to_image)
def get_gene_to_freq(path_to_file=None, db_version=None, species=None):
if path_to_file:
data = [x.strip().split('\t') for x in open(path_to_file).readlines()]
elif db_version and species:
data = [
x.strip().split('\t') for x in open(
here(PATH_TO_DB_FILES[db_version], '{}.freq.entrez.txt'.format(
species))).readlines()
]
else:
raise Exception('Specify path to file or db_version and species.')
return {int(gene): int(freq) for gene, freq in data}
def create_pretty_gmt(outpath):
print 'Collecting data'
gses = {}
for iter, m in enumerate(GQModule.objects.all()):
if iter % 1000 == 0:
print iter, 'modules done'
gse, gpl, num = m.split_full_name()
key = gse + '_' + gpl
if key not in gses:
gses[key] = {}
gses[key][num] = ToSymbolConversion.convert(
m.species, 'entrez', m.entrez_ids).get_final_symbol_ids()
# if len(gses[key][num]) != len(m.entrez_ids):
# print '{}: {} != {}'.format(key, len(gses[key][num]), len(m.entrez_ids))
print 'Writing results to', outpath
for gse in gses:
out = open(here(outpath, gse + '.gmt'), 'w')
for num in range(0, len(gses[gse])):
out.write('{}\t{}\n'.format(num, ' '.join(gses[gse][num])))
def read_queries(path):
queries = [l.strip().split('\t') for l in open(path).readlines()]
return {q[0]: set(map(int, q[-1].split())) for q in queries}
def read_modules(species):
modules = dict(
GQModule.objects.filter(species=species).values_list(
'full_name', 'entrez_ids'))
for k in modules:
modules[k] = set(modules[k])
return modules
queries = read_queries(here(GSEA_SETS_PATH, 'hs.c7.queries.txt'))
mm_gses = get_all_gses('mm')
hs_gses = get_all_gses('hs')
rt_gses = get_all_gses('rt')
hs_gene2freq = get_gene_to_freq(db_version=2015, species='hs')
mm_gene2freq = get_gene_to_freq(db_version=2015, species='mm')
rt_gene2freq = get_gene_to_freq(db_version=2015, species='rt')
hs_modules = read_modules('hs')
mm_modules = read_modules('mm')
rt_modules = read_modules('rt')
class QueryResultItem:
import os
from genequery.utils import here
ALLOWED_HOSTS = ['*']
BASE_DIR = '/genequery/front-end/genequery-web/'
DEBUG = True
STATIC_ROOT = here(BASE_DIR, '..', 'static')
STATIC_URL = '/genequery-web/static/'
STATICFILES_DIRS = (here(BASE_DIR, 'static'), )
MEDIA_ROOT = BASE_DIR + '/media/'
MEDIA_URL = '/media/'
MEDIA_ARE_LOCAL = False
# Supposed to be used for Bonferroni adjustment
# MUST BE UPDATED ALONG WITH DB VARIABLES
MODULES_COUNT = {
'hs': 117497,
'mm': 82371,
'rt': 13560,
}
BONFERRONI_ADJ_LOG_P_VALUE = {
'hs': 5.070026778096448,
'mm': 4.915774338435374,
'rt': 4.132259689531044,
}
},
'require_debug_true': {
'()': 'django.utils.log.RequireDebugTrue'
}
},
'handlers': {
'null': {
'level': 'DEBUG',
'class': 'logging.NullHandler',
},
'console': {
'level': 'DEBUG',
'class': 'logging.StreamHandler',
'formatter': 'verbose'
},
'common_file': {
'level': 'INFO',
'class': 'logging.handlers.RotatingFileHandler',
'filename': here(LOG_DIR, 'genequery.log'),
'maxBytes': 1024 * 1024 * 30,
'backupCount': 10,
'formatter': 'verbose',
},
},
'loggers': {
'genequery': {
'handlers': ['console', 'common_file'],
'level': 'DEBUG',
},
}
}