def gt(self, name, successful_ints=False, change_strint_name=False):
"""Groundtruth measures as a CausalArray.
Args:
name (str): type of groundtruth
Returns:
CausalArray: groundtruth
"""
causes = self.mutants
effects = self.genes
if name in ('gt_abs', 'gt.abs', 'abs'):
array = self.gt_abs
units = 'j'
elif name in ('gt_rel', 'gt.rel', 'rel'):
array = self.gt_rel
units = 'j/i'
elif name in ('gt_abs_norm', 'gt.abs.norm', 'abs_norm', 'abs.norm'):
array = self.gt_abs_norm
units = '1'
elif name in ('gt_abs_norm_robust', 'gt.abs.norm.robust',
'abs_norm_robust', 'abs.norm.robust'):
array = self.gt_abs_norm_robust
units = '1'
elif name in ('gt_rel_norm', 'gt.rel.norm', 'rel_norm', 'rel.norm'):
array = self.gt_rel_norm
units = '1'
elif name in ('gt_nature', 'gt.nature', 'nature'):
array = self.gt_nature
units = '1'
elif name in ('gt_sie', 'gt.sie', 'sie'):
array = self.gt_sie
units = '0'
else:
raise Exception('Unknown groundtruth %s!' % name)
## only select successfull ints, put the others to zero OR smaller
# if successful_ints:
# array[[self.map_mutants[i] for i in list(set(self.mutants).difference(set(self.successful_ints)))],:] = 0
if successful_ints:
causes = self.successful_ints
if len(causes) == 0:
raise Exception('No successful interventions found!')
array = array[[self.map_mutants[i] for i in causes], ]
if change_strint_name:
name += '_strint'
return CausalArray(array=array,
causes=causes,
effects=effects,
units=units,
name=name)
def _make_groundtruth(curated, round_size=None, verbose=True):
""" Query SGD and compile groundtruth."""
start_time = time.time()
if verbose:
print 'Querying yeastmine for %s' % 'manually curated' if curated else 'all' + 'interactions.'
# Start intermine service
yeastmine = Service('http://yeastmine.yeastgenome.org/yeastmine/service')
# 1. Get a list of all genes in quick query. Note: these are unicode formated.
query = yeastmine.new_query('Gene')
query.add_view('secondaryIdentifier')
query.add_sort_order('Gene.secondaryIdentifier', 'ASC')
query.add_constraint('Gene',
'IN',
'ALL_Verified_Uncharacterized_Dubious_ORFs',
code='A')
# create genes list and map
genes_list = []
min_length = 100
for row in query.rows():
gene = str(row['secondaryIdentifier'])
genes_list.append(str(gene))
if len(gene) < min_length:
min_length = len(gene)
if verbose: print 'Gene char length min:', min_length
genes_map = dict(zip(genes_list, range(len(genes_list))))
if verbose:
print 'Number of genes: {}'.format(
len(genes_list)
) # should be 6604 for ALL_Verified_Uncharacterized_Dubious_ORFs
# Query for interactions
query = yeastmine.new_query('Gene')
query.add_view('secondaryIdentifier', 'interactions.details.type',
'interactions.participant2.secondaryIdentifier',
'interactions.details.role1')
#query.add_view('secondaryIdentifier', 'interactions.details.type', 'interactions.details.role1')
query.add_sort_order('Gene.secondaryIdentifier', 'ASC')
query.add_constraint('organism.shortName', '=', 'S. cerevisiae', code='B')
query.add_constraint('Gene',
'IN',
'ALL_Verified_Uncharacterized_Dubious_ORFs',
code='A')
if curated:
query.add_constraint('interactions.details.annotationType',
'=',
'manually curated',
code='C')
# Matrices
adjacency_matrix_genetic = np.zeros((len(genes_list), len(genes_list)),
dtype='bool')
adjacency_matrix_physical = np.zeros((len(genes_list), len(genes_list)),
dtype='bool')
# Add to correct adjacency matrix
current_gene = None
for row in query.rows():
if (current_gene != str(row['secondaryIdentifier'])):
current_gene = str(row['secondaryIdentifier'])
print 'Querying {} - {}'.format(genes_map[current_gene],
current_gene)
# # Test genes, print all queried info
# if (current_gene == u'YAL008W'): # or current_gene == u'Q0085' or current_gene == 'Q0105'):
# if row['interactions.details.role1'] == u'Bait':
# print row['secondaryIdentifier'], '<--' , row['interactions.details.type'], '---', row['interactions.participant2.secondaryIdentifier']
# else:
# print row['secondaryIdentifier'], '---' , row['interactions.details.type'], '-->', row['interactions.participant2.secondaryIdentifier']
# Fill adjacency matrices for physical and genetic relationships
if (str(row['interactions.participant2.secondaryIdentifier'])
in genes_list):
if (row['interactions.details.type'] == u'physical interactions'):
if row['interactions.details.role1'] == u'Bait':
# Bait <--- Hit relationship found.
# maybe check if the reverse already exists?
adjacency_matrix_physical[genes_map[row['interactions.participant2.secondaryIdentifier']],\
genes_map[str(row['secondaryIdentifier'])]] = True
elif row['interactions.details.type'] == u'genetic interactions':
if row['interactions.details.role1'] == u'Bait':
# Bait <--- Hit relationship found.
adjacency_matrix_genetic[genes_map[row['interactions.participant2.secondaryIdentifier']],\
genes_map[str(row['secondaryIdentifier'])]] = True
if verbose:
print 'Nuber of genetic interactions {}'.format(
adjacency_matrix_genetic.sum())
print 'Nuber of physical interactions {}'.format(
adjacency_matrix_physical.sum())
# save all data + write date in each.
CausalArray(array=adjacency_matrix_genetic,
causes=genes_list,
name='sgd_gen{}'.format('_cur' if curated else '')).save(
file=filename_groundtruth.format(
'_curated' if curated else '', 'genetic'))
# write_datestamp_hdf5(filename_groundtruth.format('_curated' if curated else '', 'genetic'))
CausalArray(array=adjacency_matrix_physical,
causes=genes_list,
name='sgd_phys{}'.format('_cur' if curated else '')).save(
file=filename_groundtruth.format(
'_curated' if curated else '', 'physical'))
# write_datestamp_hdf5(filename_groundtruth.format('_curated' if curated else '', 'physical'))
CausalArray(array=np.add(adjacency_matrix_genetic,
adjacency_matrix_physical),
causes=genes_list,
name='sgd_all{}'.format('_cur' if curated else '')).save(
file=filename_groundtruth.format(
'_curated' if curated else '', 'all'))
# write_datestamp_hdf5(filename_groundtruth.format('_curated' if curated else '', 'all'))
if verbose: print 'Done in {:.2f}'.format(time.time() - start_time)