def find_saschas_pairs():
pair_dists = biopsy.pair_analysis(
'c:/data/remos/0/0.cleaned',
'c:/data/remos/0/0.analysis',
45,
pair_filter = saschas_pair_filter,
location_filter = biopsy.get_default_pair_filter(
family_file = 'C:/Data/Ensembl/mouse_families.txt',
paralog_file = 'C:/Data/Ensembl/mouse_paralogs.txt',
subtypes = [
# 'Fungi/Metazoa group',
# 'Coelomata',
# 'Bilateria',
# 'Chordata',
# 'Amniota',
# 'Tetrapoda',
'Euteleostomi',
'Theria',
'Eutheria',
'Euarchontoglires',
'Murinae',
'Mus musculus',
]
)
)
biopsy.save( pair_dists, 'saschas.pairs' )
def test_pair_analysis():
import webbrowser
remome_threshold = 100
remome_file = 'c:/data/remos/%d/%d.cleaned' % ( remome_threshold, remome_threshold )
analysis_file = 'c:/data/remos/%d/%d.analysis' % ( remome_threshold, remome_threshold )
location_filter = biopsy.get_default_pair_filter()
hits = biopsy.pair_analysis(
remome_file,
analysis_file,
max_distance = 45,
distance_filter = lambda d: d == 25,
pair_filter = lambda p: p == ( 'M00471', 'M00791', False, False ),
location_filter = location_filter
)
for k, v in hits.iteritems():
print k
print v
for g in location_filter.filter2.genes_for_pair[ k ]:
print g
webbrowser.open( biopsy.ensembl_url( g ) )
print