def convert_paragraph(new_session_maker):
from model_new_schema.bioentity import Bioentity, Paragraph
log = logging.getLogger('convert.regulation.paragraph')
log.info('begin')
output_creator = OutputCreator(log)
try:
new_session = new_session_maker()
#Values to check
values_to_check = ['text', 'date_created', 'created_by']
#Grab cached dictionaries
key_to_bioentity = dict([(x.unique_key(), x) for x in new_session.query(Bioentity).all()])
#Grab all current objects
current_objs = new_session.query(Paragraph).filter(Paragraph.class_type == 'REGULATION').all()
id_to_current_obj = dict([(x.id, x) for x in current_objs])
key_to_current_obj = dict([(x.unique_key(), x) for x in current_objs])
untouched_obj_ids = set(id_to_current_obj.keys())
old_objs = break_up_file('/Users/kpaskov/final/Reg_Summary_Paragraphs04282013.txt')
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_paragraph(old_obj, key_to_bioentity)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = newly_created_obj.unique_key()
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if unique_key not in key_to_current_obj else key_to_current_obj[unique_key]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
#Delete untouched objs
for untouched_obj_id in untouched_obj_ids:
new_session.delete(id_to_current_obj[untouched_obj_id])
output_creator.removed()
#Commit
output_creator.finished()
new_session.commit()
except Exception:
log.exception('Unexpected error:' + str(sys.exc_info()[0]))
finally:
new_session.close()
log.info('complete')
def convert_evidence(new_session_maker, chunk_size):
from model_new_schema.regulation import Regulationevidence
from model_new_schema.evelement import Experiment
from model_new_schema.bioentity import Bioentity
from model_new_schema.reference import Reference
log = logging.getLogger('convert.regulation.evidence')
log.info('begin')
output_creator = OutputCreator(log)
try:
new_session = new_session_maker()
#Values to check
values_to_check = ['experiment_id', 'reference_id', 'strain_id', 'source', 'conditions',
'bioentity1_id', 'bioentity2_id', 'date_created', 'created_by']
#Grab cached dictionaries
key_to_experiment = dict([(x.unique_key(), x) for x in new_session.query(Experiment).all()])
key_to_bioent = dict([(x.unique_key(), x) for x in new_session.query(Bioentity).all()])
pubmed_to_reference_id = dict([(x.pubmed_id, x.id) for x in new_session.query(Reference).all()])
#Grab old objects
data = break_up_file('/Users/kpaskov/final/yeastmine_regulation.tsv')
count = len(data)
num_chunks = ceil(1.0*count/chunk_size)
min_id = 0
j = 0
for i in range(0, num_chunks):
#Grab all current objects
current_objs = new_session.query(Regulationevidence).filter(Regulationevidence.id >= create_evidence_id(min_id)).filter(Regulationevidence.id < create_evidence_id(min_id+chunk_size)).all()
id_to_current_obj = dict([(x.id, x) for x in current_objs])
key_to_current_obj = dict([(x.unique_key(), x) for x in current_objs])
untouched_obj_ids = set(id_to_current_obj.keys())
old_objs = data[min_id:min_id+chunk_size]
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_evidence(old_obj, j, key_to_experiment, key_to_bioent, pubmed_to_reference_id)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = newly_created_obj.unique_key()
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if unique_key not in key_to_current_obj else key_to_current_obj[unique_key]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
j = j + 1
#Delete untouched objs
for untouched_obj_id in untouched_obj_ids:
new_session.delete(id_to_current_obj[untouched_obj_id])
output_creator.removed()
output_creator.finished(str(i+1) + "/" + str(int(num_chunks)))
new_session.commit()
min_id = min_id+chunk_size
#Commit
output_creator.finished()
new_session.commit()
except Exception:
log.exception('Unexpected error:' + str(sys.exc_info()[0]))
finally:
new_session.close()
log.info('complete')
def convert_domain(new_session_maker, chunk_size):
from model_new_schema.protein import Domain as Domain
log = logging.getLogger('convert.protein.domain')
log.info('begin')
output_creator = OutputCreator(log)
try:
#Grab all current objects
new_session = new_session_maker()
current_objs = new_session.query(Domain).all()
id_to_current_obj = dict([(x.id, x) for x in current_objs])
key_to_current_obj = dict([(x.unique_key(), x) for x in current_objs])
#Values to check
values_to_check = ['display_name', 'description', 'interpro_id', 'interpro_description', 'link']
untouched_obj_ids = set(id_to_current_obj.keys())
#Grab old objects
data = break_up_file('/Users/kpaskov/final/yeastmine_protein_domains.tsv')
used_unique_keys = set()
min_id = 0
count = len(data)
num_chunks = ceil(1.0*count/chunk_size)
for i in range(0, num_chunks):
old_objs = data[min_id:min_id+chunk_size]
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_domain(old_obj)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = newly_created_obj.unique_key()
if unique_key not in used_unique_keys:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if unique_key not in key_to_current_obj else key_to_current_obj[unique_key]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
used_unique_keys.add(unique_key)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
output_creator.finished(str(i+1) + "/" + str(int(num_chunks)))
new_session.commit()
min_id = min_id+chunk_size
#Grab JASPAR domains from file
old_objs = break_up_file('/Users/kpaskov/final/TF_family_class_accession04302013.txt')
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_domain_from_tf_file(old_obj)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = newly_created_obj.unique_key()
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if unique_key not in key_to_current_obj else key_to_current_obj[unique_key]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
used_unique_keys.add(unique_key)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
output_creator.finished("1/1")
new_session.commit()
#Delete untouched objs
for untouched_obj_id in untouched_obj_ids:
new_session.delete(id_to_current_obj[untouched_obj_id])
output_creator.removed()
#Commit
output_creator.finished()
new_session.commit()
except Exception:
log.exception('Unexpected error:' + str(sys.exc_info()[0]))
finally:
new_session.close()
log.info('complete')
def convert_experiment(old_session_maker, new_session_maker):
from model_new_schema.evelement import Experiment as NewExperiment
from model_old_schema.cv import CVTerm as OldCVTerm
log = logging.getLogger('convert.evelements.experiment')
log.info('begin')
output_creator = OutputCreator(log)
try:
#Grab all current objects
new_session = new_session_maker()
current_objs = new_session.query(NewExperiment).all()
id_to_current_obj = dict([(x.id, x) for x in current_objs])
key_to_current_obj = dict([(x.unique_key(), x) for x in current_objs])
#Values to check
values_to_check = ['display_name', 'link', 'description', 'date_created', 'created_by', 'eco_id']
untouched_obj_ids = set(id_to_current_obj.keys())
#Grab old objects
old_session = old_session_maker()
old_objs = old_session.query(OldCVTerm).filter(OldCVTerm.cv_no==7).all()
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_experiment(old_obj)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if newly_created_obj.unique_key() not in key_to_current_obj else key_to_current_obj[newly_created_obj.unique_key()]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
#Get experiments from regulation files
experiment_names = set()
rows = break_up_file('/Users/kpaskov/final/yeastmine_regulation.tsv')
experiment_names.update([(row[4], row[5]) for row in rows])
i=0
for experiment_name, eco_id in experiment_names:
newly_created_objs = create_experiment_from_reg_row(experiment_name, eco_id, i)
for newly_created_obj in newly_created_objs:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if newly_created_obj.unique_key() not in key_to_current_obj else key_to_current_obj[newly_created_obj.unique_key()]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
i = i+1
experiment_names = set()
#Add experiments from binding files
rows = break_up_file('/Users/kpaskov/final/yetfasco_data.txt', delimeter=';')
for row in rows:
if len(row) < 10:
print row
experiment_names.update([row[9][1:-1] for row in rows])
i=0
for experiment_name in experiment_names:
newly_created_objs = create_experiment_from_binding_row(experiment_name, i)
for newly_created_obj in newly_created_objs:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if newly_created_obj.unique_key() not in key_to_current_obj else key_to_current_obj[newly_created_obj.unique_key()]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
i = i+1
#Delete untouched objs
for untouched_obj_id in untouched_obj_ids:
new_session.delete(id_to_current_obj[untouched_obj_id])
print 'Removed at end'
output_creator.removed()
#Commit
output_creator.finished()
new_session.commit()
except Exception:
log.exception('Unexpected error:' + str(sys.exc_info()[0]))
finally:
new_session.close()
old_session.close()
log.info('complete')
def convert_domain_evidence(new_session_maker, chunk_size):
from model_new_schema.protein import Domain, Domainevidence
from model_new_schema.bioentity import Bioentity
from model_new_schema.reference import Reference
log = logging.getLogger('convert.protein.domain_evidence')
log.info('begin')
output_creator = OutputCreator(log)
try:
#Grab all current objects
new_session = new_session_maker()
current_objs = new_session.query(Domainevidence).all()
id_to_current_obj = dict([(x.id, x) for x in current_objs])
key_to_current_obj = dict([(x.unique_key(), x) for x in current_objs])
#Values to check
values_to_check = ['reference_id', 'strain_id', 'source', 'date_created', 'created_by',
'start', 'end', 'evalue', 'status', 'date_of_run', 'protein_id', 'domain_id']
#Grab cached dictionaries
key_to_bioentity = dict([(x.unique_key(), x) for x in new_session.query(Bioentity).all()])
key_to_domain = dict([(x.unique_key(), x) for x in new_session.query(Domain).all()])
pubmed_id_to_reference_id = dict([(x.pubmed_id, x.id) for x in new_session.query(Reference).all()])
untouched_obj_ids = set(id_to_current_obj.keys())
#Grab old objects
data = break_up_file('/Users/kpaskov/final/yeastmine_protein_domains.tsv')
used_unique_keys = set()
j=0
min_id = 0
count = len(data)
num_chunks = ceil(1.0*count/chunk_size)
for i in range(0, num_chunks):
old_objs = data[min_id:min_id+chunk_size]
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_domain_evidence(old_obj, j, key_to_bioentity, key_to_domain)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = (newly_created_obj.protein_id, newly_created_obj.domain_id, newly_created_obj.start,
newly_created_obj.end, newly_created_obj.evalue)
if unique_key not in used_unique_keys:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if newly_created_obj.id not in key_to_current_obj else key_to_current_obj[newly_created_obj.id]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
used_unique_keys.add(unique_key)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
j = j+1
output_creator.finished(str(i+1) + "/" + str(int(num_chunks)))
new_session.commit()
min_id = min_id+chunk_size
#Grab JASPAR evidence from file
old_objs = break_up_file('/Users/kpaskov/final/TF_family_class_accession04302013.txt')
for old_obj in old_objs:
#Convert old objects into new ones
newly_created_objs = create_domain_evidence_from_tf_file(old_obj, j, key_to_bioentity, key_to_domain, pubmed_id_to_reference_id)
if newly_created_objs is not None:
#Edit or add new objects
for newly_created_obj in newly_created_objs:
unique_key = newly_created_obj.unique_key()
if unique_key not in used_unique_keys:
current_obj_by_id = None if newly_created_obj.id not in id_to_current_obj else id_to_current_obj[newly_created_obj.id]
current_obj_by_key = None if unique_key not in key_to_current_obj else key_to_current_obj[unique_key]
create_or_update(newly_created_obj, current_obj_by_id, current_obj_by_key, values_to_check, new_session, output_creator)
used_unique_keys.add(unique_key)
if current_obj_by_id is not None and current_obj_by_id.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_id.id)
if current_obj_by_key is not None and current_obj_by_key.id in untouched_obj_ids:
untouched_obj_ids.remove(current_obj_by_key.id)
j = j+1
output_creator.finished("1/1")
new_session.commit()
#Delete untouched objs
for untouched_obj_id in untouched_obj_ids:
new_session.delete(id_to_current_obj[untouched_obj_id])
output_creator.removed()
#Commit
output_creator.finished()
new_session.commit()
except Exception:
log.exception('Unexpected error:' + str(sys.exc_info()[0]))
finally:
new_session.close()
log.info('complete')
