def main_func(self, positions, iteration):
# filenames
if not positions[1]:
rows = self.data[positions[0]:]
else:
rows = self.data[positions[0]:positions[1]]
missing_proteins = {}
mutants_for_proteins = {}
c = 0
skipped = 0
inserted = 0
bulk_m = []
bulk_r = []
current_sheet = time.time()
for r in rows:
# print(source_file,c)
# PRINT IF ERRORS OCCUR
# self.logger.info('File '+str(r['source_file'])+' number '+str(c))
current = time.time()
c += 1
# if c%100==0:
# self.logger.info('Parsed '+str(c)+' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
float(r['review'])
r['review'] = str(int(r['review']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
if r['reference'] not in self.publication_cache:
try:
pub = Publication.objects.get(
web_link__index=r['reference'],
web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = WebLink.objects.get(
index=r['reference'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(
index=r['reference'],
web_resource=WebResource.objects.get(
slug=pub_type))
pub.web_link = wl
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' +
str(r['reference']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['reference']] = pub
else:
pub = self.publication_cache[r['reference']]
# print(r['review'],r['reference'])
if r['review'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
# print(r['review'],pub_type)
if r['review']:
if r['review'] not in self.publication_cache:
try:
pub_review = Publication.objects.get(
web_link__index=r['review'],
web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub_review = Publication()
try:
pub_review.web_link = WebLink.objects.get(
index=r['review'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(
index=r['review'],
web_resource=WebResource.objects.get(
slug=pub_type))
pub_review.web_link = wl
if pub_type == 'doi':
pub_review.update_from_doi(doi=r['review'])
elif pub_type == 'pubmed':
pub_review.update_from_pubmed_data(
index=r['review'])
try:
pub_review.save()
except:
self.logger.error('error with review ' +
str(r['review']) + ' ' +
pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['review']] = pub_review
else:
pub_review = self.publication_cache[r['review']]
else:
pub_review = None
l = None
if str(r['ligand_name']) in self.ligand_cache:
if r['ligand_id'] in self.ligand_cache[str(r['ligand_name'])]:
l = self.ligand_cache[str(
r['ligand_name'])][r['ligand_id']]
else:
self.ligand_cache[str(r['ligand_name'])] = {}
if not l:
l = get_or_make_ligand(r['ligand_id'], r['ligand_type'],
str(r['ligand_name']))
self.ligand_cache[str(r['ligand_name'])][r['ligand_id']] = l
l_ref = None
if str(r['exp_mu_ligand_ref']) in self.ref_ligand_cache:
l_ref = self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])]
else:
if Ligand.objects.filter(
name=r['exp_mu_ligand_ref'], canonical=True
).exists(
): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=True)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False
).exists(
): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=True
).exists(
): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'],
canonical=False).exists(
): #amigious_alias not specified
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False)
l_ref.ambigious_alias = False
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
try:
l_ref.save()
except IntegrityError:
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False)
print("error failing ligand, duplicate?")
# logger.error("FAILED SAVING LIGAND, duplicate?")
else:
l_ref = None
self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])] = l_ref
protein_id = 0
residue_id = 0
protein = Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein = protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
self.logger.error('Skipped due to no protein ' +
r['protein'])
continue
res = Residue.objects.filter(
protein_conformation__protein=protein,
amino_acid=r['mutation_from'],
sequence_number=r['mutation_pos']) #FIXME MAKE AA CHECK
if res.exists():
res = res.get()
else:
self.logger.error('Skipped due to no residue or mismatch AA ' +
r['protein'] + ' pos:' +
str(r['mutation_pos']) + ' AA:' +
r['mutation_from'])
skipped += 1
continue
if r['ligand_class']:
l_role, created = LigandRole.objects.get_or_create(
name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]
}) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(
type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(
func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r['exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(
qual=r['exp_mu_effect_qual'],
prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r[
'opt_sign'] or r['opt_percentage'] or r['opt_qual'] or r[
'opt_agonist']:
exp_opt_id, created = MutationOptional.objects.get_or_create(
type=r['opt_type'],
wt=r['opt_wt'],
mu=r['opt_mu'],
sign=r['opt_sign'],
percentage=r['opt_percentage'],
qual=r['opt_qual'],
agonist=r['opt_agonist'])
else:
exp_opt_id = None
try:
mutation, created = Mutation.objects.get_or_create(
amino_acid=r['mutation_to'], protein=protein, residue=res)
except IntegrityError:
mutation = Mutation.objects.get(amino_acid=r['mutation_to'],
protein=protein,
residue=res)
logtypes = ['pEC50', 'pIC50', 'pK']
foldchange = 0
typefold = ''
if r['exp_wt_value'] != 0 and r[
'exp_mu_value_raw'] != 0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")",
r['exp_type']): #-log values!
foldchange = round(
math.pow(10, -r['exp_mu_value_raw']) /
pow(10, -r['exp_wt_value']), 3)
typefold = r['exp_type'] + "_log"
else:
foldchange = round(
r['exp_mu_value_raw'] / r['exp_wt_value'], 3)
typefold = r['exp_type'] + "_not_log"
if foldchange < 1 and foldchange != 0:
foldchange = -round((1 / foldchange), 3)
elif r['fold_effect'] != 0:
foldchange = round(r['fold_effect'], 3)
if foldchange < 1: foldchange = -round((1 / foldchange), 3)
raw_experiment = self.insert_raw(r)
bulk = MutationExperiment(
refs=pub,
review=pub_review,
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref=l_ref,
#raw = raw_experiment, #raw_experiment, OR None
optional=exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual=exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value=r['exp_mu_value_raw'],
mu_sign=r['exp_mu_effect_sign'],
foldchange=foldchange)
# mut_id = obj.id
bulk_r.append(raw_experiment)
bulk_m.append(bulk)
inserted += 1
end = time.time()
diff = round(end - current, 2)
#print(diff)
self.logger.info('Parsed ' + str(c) +
' mutant data entries. Skipped ' + str(skipped))
current = time.time()
raws = MutationRaw.objects.bulk_create(bulk_r)
for i, me in enumerate(bulk_m):
me.raw = raws[i]
MutationExperiment.objects.bulk_create(bulk_m)
end = time.time()
diff = round(end - current, 2)
current_sheet
diff_2 = round(end - current_sheet, 2)
print("overall", diff_2, "bulk", diff, len(bulk_m), "skipped",
str(skipped))
def get_or_make_ligand(ligand_id, type_id, name=None):
if type_id == 'PubChem CID' or type_id == 'SMILES':
if type_id == 'PubChem CID':
pubchem_lookup_value = 'cid'
elif type_id == 'SMILES':
pubchem_lookup_value = 'smiles'
try:
web_resource = WebResource.objects.get(slug='pubchem')
except:
# abort if pdb resource is not found
raise Exception('PubChem resource not found, aborting!')
if name:
ligand_name = name
else:
ligand_name = False
try:
# if this name is canonical and it has a ligand record already
if (ligand_name == False):
l = None
ls = Ligand.objects.filter(
canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
for ligand in ls:
l = ligand
#print (l)
break
if l == None:
l = Ligand.objects.get(
canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
else:
l = Ligand.objects.get(
name=ligand_name,
canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
#l = Ligand.objects.get(name=ligand_name, canonical=True,
# properities__web_links__web_resource=web_resource,
# properities__web_links__index=ligand_id)
#
except Ligand.DoesNotExist:
try:
# if exists under different name
l_canonical = Ligand.objects.get(
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id,
canonical=True)
#print (created)
try:
l, created = Ligand.objects.get_or_create(
properities=l_canonical.properities,
name=ligand_name,
canonical=False)
except IntegrityError:
l = Ligand.objects.get(properities=l_canonical.properities,
name=ligand_name,
canonical=False)
except Ligand.DoesNotExist:
# fetch ligand from pubchem
default_ligand_type = 'Small molecule'
lt, created = LigandType.objects.get_or_create(
slug=slugify(default_ligand_type),
defaults={'name': default_ligand_type})
l = Ligand()
#print (ligand_name)
l = l.load_from_pubchem(pubchem_lookup_value, ligand_id, lt,
ligand_name)
#print (l)
if l == None and type_id == 'SMILES': #insert manually if smiles and unfound in pubchem
try:
l = Ligand.objects.get(name=ligand_name,
canonical=True,
properities__smiles=ligand_id)
except Ligand.DoesNotExist:
try:
l = Ligand.objects.get(
name__startswith=ligand_name,
canonical=True,
properities__smiles=ligand_id
) #if no properities exist
except Ligand.DoesNotExist:
try:
l = Ligand.objects.get(
name=ligand_name,
canonical=True,
properities__smiles=None
) #if no properities exist
l.properities.smiles = ligand_id
l.properities.save()
l.save()
except Ligand.DoesNotExist:
## now insert a new ligand, but first make sure name is unique
if Ligand.objects.filter(
name=ligand_name).exists():
ls = Ligand.objects.filter(
name__startswith=ligand_name,
canonical=True).order_by("pk")
for l_temp in ls:
last = l_temp.name.split("_")[-1]
if last == ligand_name: #no addition yet
ligand_name = ligand_name + "_1"
else:
ligand_name = ligand_name + "_" + str(
int(last) + 1)
l = Ligand()
l.name = ligand_name
lp = LigandProperities()
lp.smiles = ligand_id
lp.ligand_type = lt
lp.save()
l.properities = lp
l.canonical = True #maybe false, but that would break stuff.
l.ambigious_alias = False
try:
l.save()
except IntegrityError:
l = Ligand.objects.get(name=ligand_name,
canonical=True)
elif name:
# if this name is canonical and it has a ligand record already
if Ligand.objects.filter(name=name, canonical=True).exists():
l = Ligand.objects.get(name=name, canonical=True)
# if this matches an alias that only has "one" parent canonical name - eg distinct
elif Ligand.objects.filter(name=name,
canonical=False,
ambigious_alias=False).exists():
l = Ligand.objects.get(name=name,
canonical=False,
ambigious_alias=False)
# if this matches an alias that only has several canonical parents, must investigate, start
# with empty.
elif Ligand.objects.filter(name=name,
canonical=False,
ambigious_alias=True).exists():
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = name
l.canonical = False
l.ambigious_alias = True
l.save()
l.load_by_name(name)
# if neither a canonical or alias exists, create the records. Remember to check for
# canonical / alias status.
else:
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = str(name)
l.canonical = True
l.ambigious_alias = False
try:
l.save()
l.load_by_name(str(name))
except IntegrityError:
l = Ligand.objects.get(name=str(name), canonical=True)
else:
l = None
return l
def create_mutant_data(self, filenames):
self.logger.info('CREATING MUTANT DATA')
# what files should be parsed?
if not filenames:
filenames = os.listdir(self.structure_data_dir)
missing_proteins = {}
mutants_for_proteins = {}
for source_file in filenames:
source_file_path = os.sep.join([self.structure_data_dir, source_file])
if os.path.isfile(source_file_path) and source_file[0] != '.':
self.logger.info('Reading file {}'.format(source_file_path))
# read the yaml file
if source_file[-4:]=='xlsx' or source_file[-3:]=='xls':
rows = self.loaddatafromexcel(source_file_path)
rows = self.analyse_rows(rows)
elif source_file[-4:]=='yaml':
rows = yaml.load(open(source_file_path, 'r'))
temp = []
for r in rows:
d = {}
d['reference'] = r['pubmed']
d['protein'] = r['entry_name'].replace("__","_").lower()
d['mutation_pos'] = r['seq']
d['mutation_from'] = r['from_res']
d['mutation_to'] = r['to_res']
d['ligand_name'] = ''
d['ligand_type'] = ''
d['ligand_id'] = ''
d['ligand_class'] = ''
d['exp_type'] = ''
d['exp_func'] = ''
d['exp_wt_value'] = 0
d['exp_wt_unit'] = ''
d['exp_mu_effect_sign'] = ''
d['exp_mu_value_raw'] = 0
d['fold_effect'] = 0
d['exp_mu_effect_qual'] = ''
d['exp_mu_effect_ligand_prop'] = ''
d['exp_mu_ligand_ref'] = ''
d['opt_type'] = ''
d['opt_wt'] = 0
d['opt_mu'] = 0
d['opt_sign'] = ''
d['opt_percentage'] = 0
d['opt_qual'] = ''
d['opt_agonist'] = ''
if len(d['mutation_to'])>1 or len(d['mutation_from'])>1: #if something is off with amino acid
continue
temp.append(d)
rows = temp
else:
self.logger.info('unknown format'.source_file)
continue
c = 0
skipped = 0
inserted = 0
for r in rows:
c += 1
if c%1000==0:
self.logger.info('Parsed '+str(c)+' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
try:
pub = Publication.objects.get(web_link__index=r['reference'], web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = WebLink.objects.get(index=r['reference'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(index=r['reference'],
web_resource = WebResource.objects.get(slug=pub_type))
pub.web_link = wl
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' + str(r['reference']) + ' ' + pub_type)
continue #if something off with publication, skip.
if r['ligand_type']=='PubChem CID' or r['ligand_type']=='SMILES':
if r['ligand_type']=='PubChem CID':
pubchem_lookup_value = 'cid'
elif r['ligand_type']=='SMILES':
pubchem_lookup_value = 'smiles'
try:
web_resource = WebResource.objects.get(slug='pubchem')
except:
# abort if pdb resource is not found
raise Exception('PubChem resource not found, aborting!')
if 'ligand_name' in r and r['ligand_name']:
ligand_name = str(r['ligand_name'])
else:
ligand_name = False
try:
# if this name is canonical and it has a ligand record already
l = Ligand.objects.get(name=ligand_name, canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=r['ligand_id'])
except Ligand.DoesNotExist:
try:
# if exists under different name
l_canonical = Ligand.objects.get(properities__web_links__web_resource=web_resource,
properities__web_links__index=r['ligand_id'], canonical=True)
l, created = Ligand.objects.get_or_create(properities = l_canonical.properities,
name = ligand_name, canonical = False)
if created:
self.logger.info('Created ligand {}'.format(l.name))
except Ligand.DoesNotExist:
# fetch ligand from pubchem
default_ligand_type = 'Small molecule'
lt, created = LigandType.objects.get_or_create(slug=slugify(default_ligand_type),
defaults={'name': default_ligand_type})
l = Ligand()
l = l.load_from_pubchem(pubchem_lookup_value, r['ligand_id'], lt, ligand_name)
elif r['ligand_name']:
# if this name is canonical and it has a ligand record already
if Ligand.objects.filter(name=r['ligand_name'], canonical=True).exists():
l = Ligand.objects.get(name=r['ligand_name'], canonical=True)
# if this matches an alias that only has "one" parent canonical name - eg distinct
elif Ligand.objects.filter(name=r['ligand_name'], canonical=False,
ambigious_alias=False).exists():
l = Ligand.objects.get(name=r['ligand_name'], canonical=False, ambigious_alias=False)
# if this matches an alias that only has several canonical parents, must investigate, start
# with empty.
elif Ligand.objects.filter(name=r['ligand_name'], canonical=False,
ambigious_alias=True).exists():
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = r['ligand_name']
l.canonical = False
l.ambigious_alias = True
l.save()
l.load_by_name(r['ligand_name'])
# if neither a canonical or alias exists, create the records. Remember to check for
# canonical / alias status.
else:
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = str(r['ligand_name'])
l.canonical = True
l.ambigious_alias = False
l.save()
l.load_by_name(str(r['ligand_name']))
else:
l = None
if Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=True).exists(): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=True)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False).exists(): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=True).exists(): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
else:
l_ref = None
protein_id = 0
residue_id = 0
protein=Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein=protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
self.logger.error('Skipped due to no protein '+ r['protein'])
continue
res=Residue.objects.filter(protein_conformation__protein=protein,sequence_number=r['mutation_pos'])
if res.exists():
res=res.get()
else:
self.logger.error('Skipped due to no residue ' + r['protein'] + ' pos:'+str(r['mutation_pos']))
skipped += 1
continue
if r['ligand_class']:
l_role, created = LigandRole.objects.get_or_create(name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]}) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r['exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(qual=r['exp_mu_effect_qual'], prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r['opt_sign'] or r['opt_percentage'] or r['opt_qual'] or r['opt_agonist']:
exp_opt_id, created = MutationOptional.objects.get_or_create(type=r['opt_type'], wt=r['opt_wt'], mu=r['opt_mu'], sign=r['opt_sign'], percentage=r['opt_percentage'], qual=r['opt_qual'], agonist=r['opt_agonist'])
else:
exp_opt_id = None
mutation, created = Mutation.objects.get_or_create(amino_acid=r['mutation_to'],protein=protein, residue=res)
logtypes = ['pEC50','pIC50','pK']
foldchange = 0
typefold = ''
if r['exp_wt_value']!=0 and r['exp_mu_value_raw']!=0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")", r['exp_type']): #-log values!
foldchange = round(math.pow(10,-r['exp_mu_value_raw'])/pow(10,-r['exp_wt_value']),3);
typefold = r['exp_type']+"_log"
else:
foldchange = round(r['exp_mu_value_raw']/r['exp_wt_value'],3);
typefold = r['exp_type']+"_not_log"
if foldchange<1 and foldchange!=0:
foldchange = -round((1/foldchange),3)
elif r['fold_effect']!=0:
foldchange = round(r['fold_effect'],3);
if foldchange<1: foldchange = -round((1/foldchange),3);
raw_experiment = self.insert_raw(r)
obj, created = MutationExperiment.objects.get_or_create(
refs=pub,
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref = l_ref,
raw = raw_experiment,
optional = exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual = exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value = r['exp_mu_value_raw'],
mu_sign = r['exp_mu_effect_sign'],
foldchange = foldchange
)
mut_id = obj.id
inserted += 1
self.logger.info('Parsed '+str(c)+' mutant data entries. Skipped '+str(skipped))
sorted_missing_proteins = sorted(missing_proteins.items(), key=operator.itemgetter(1),reverse=True)
sorted_mutants_for_proteins = sorted(mutants_for_proteins.items(), key=operator.itemgetter(1),reverse=True)
self.logger.info('COMPLETED CREATING MUTANTS')
def main_func(self, positions, iteration, count, lock):
# filenames
# if not positions[1]:
# rows = self.data[positions[0]:]
# else:
# rows = self.data[positions[0]:positions[1]]
missing_proteins = {}
mutants_for_proteins = {}
wrong_uniport_ids = {}
c = 0
skipped = 0
inserted = 0
bulk_m = []
bulk_r = []
current_sheet = time.time()
rows = self.data_all
while count.value < len(rows):
with lock:
r = rows[count.value]
count.value += 1
# for r in rows:
# print(r['source_file'],c)
# PRINT IF ERRORS OCCUR
#self.logger.info('File '+str(r['source_file'])+' number '+str(c))
current = time.time()
c += 1
# if c%100==0:
# self.logger.info('Parsed '+str(c)+' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
float(r['review'])
r['review'] = str(int(r['review']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
if r['reference'] not in self.publication_cache:
try:
wl = WebLink.objects.get(index=r['reference'],
web_resource__slug=pub_type)
except WebLink.DoesNotExist:
try:
wl = WebLink.objects.create(
index=r['reference'],
web_resource=WebResource.objects.get(
slug=pub_type))
except IntegrityError:
wl = WebLink.objects.get(index=r['reference'],
web_resource__slug=pub_type)
try:
pub = Publication.objects.get(web_link=wl)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = wl
pub.save()
except IntegrityError:
pub = Publication.objects.get(web_link=wl)
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' +
str(r['reference']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['reference']] = pub
else:
pub = self.publication_cache[r['reference']]
# print(r['review'],r['reference'])
if r['review'].isdigit(): #assume pubmed
pub_type = 'pubmed'
elif r['review'].startswith('http'):
pub_type = 'raw_link'
else: #assume doi
pub_type = 'doi'
# print(r['review'],pub_type)
if r['review']:
if r['review'] not in self.publication_cache:
try:
wl = WebLink.objects.get(index=r['review'],
web_resource__slug=pub_type)
except WebLink.DoesNotExist:
try:
wl = WebLink.objects.create(
index=r['review'],
web_resource=WebResource.objects.get(
slug=pub_type))
except IntegrityError:
wl = WebLink.objects.get(
index=r['review'], web_resource__slug=pub_type)
try:
pub_review = Publication.objects.get(web_link=wl)
except Publication.DoesNotExist:
pub_review = Publication()
try:
pub_review.web_link = wl
pub_review.save()
except IntegrityError:
pub_review = Publication.objects.get(web_link=wl)
if pub_type == 'doi':
pub_review.update_from_doi(doi=r['review'])
elif pub_type == 'pubmed':
pub_review.update_from_pubmed_data(
index=r['review'])
try:
pub_review.save()
except:
self.logger.error('error with review ' +
str(r['review']) + ' ' +
pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['review']] = pub_review
else:
pub_review = self.publication_cache[r['review']]
else:
pub_review = None
l = None
if str(r['ligand_name']) in self.ligand_cache:
if r['ligand_id'] in self.ligand_cache[str(r['ligand_name'])]:
l = self.ligand_cache[str(
r['ligand_name'])][r['ligand_id']]
else:
self.ligand_cache[str(r['ligand_name'])] = {}
if not l:
try:
l = get_or_make_ligand(r['ligand_id'], r['ligand_type'],
str(r['ligand_name']))
except Exception as msg:
print(
'Something errored with ligand, aborting entry of mutation',
r['ligand_name'], r['ligand_type'], r['ligand_id'],
r['source_file'])
print(msg)
traceback.print_exc()
continue
self.ligand_cache[str(r['ligand_name'])][r['ligand_id']] = l
l_ref = None
if str(r['exp_mu_ligand_ref']) in self.ref_ligand_cache:
l_ref = self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])]
else:
if Ligand.objects.filter(
name=r['exp_mu_ligand_ref'], canonical=True
).exists(
): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=True)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False
).exists(
): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=True
).exists(
): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'],
canonical=False).exists(
): #amigious_alias not specified
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False)
l_ref.ambigious_alias = False
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
try:
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
except IntegrityError:
if Ligand.objects.filter(name=r['exp_mu_ligand_ref'],
canonical=True).exists():
l_ref = Ligand.objects.get(
name=r['exp_mu_ligand_ref'], canonical=True)
else:
l_ref = Ligand.objects.get(
name=r['exp_mu_ligand_ref'], canonical=False)
# print("error failing ligand, duplicate?")
try:
l_ref.save()
except IntegrityError:
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False)
# print("error failing ligand, duplicate?")
# logger.error("FAILED SAVING LIGAND, duplicate?")
else:
l_ref = None
self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])] = l_ref
protein_id = 0
residue_id = 0
protein = Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein = protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
elif r['protein'] not in missing_proteins:
try:
r['protein'] = wrong_uniport_ids[r['protein']]
real_uniprot = wrong_uniport_ids[r['protein']]
protein = Protein.objects.get(entry_name=r['protein'])
# print('fetched with lookup table',r['protein'])
except:
# look for it as uniprot
protein = Protein.objects.filter(
web_links__web_resource__slug='uniprot',
web_links__index=r['protein'].upper())
if protein.exists():
protein = protein.get()
real_uniprot = protein.entry_name
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
# Try to lookup in uniprot to catch typing errors / variants in entry_name
url = 'http://www.uniprot.org/uniprot/$index.xml'
cache_dir = ['uniprot', 'id']
uniprot_protein = fetch_from_web_api(url,
r['protein'],
cache_dir,
xml=True)
try:
real_uniprot = uniprot_protein.find(
'.//{http://uniprot.org/uniprot}name'
).text.lower()
protein = Protein.objects.get(
entry_name=real_uniprot)
except:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
# print('Skipped due to no protein '+ r['protein'])
self.logger.error(
'Skipped due to no protein ' +
r['protein'])
continue
wrong_uniport_ids[r['protein']] = protein.entry_name
r['protein'] = real_uniprot
else:
missing_proteins[r['protein']] += 1
continue
res = Residue.objects.filter(
protein_conformation__protein=protein,
amino_acid=r['mutation_from'],
sequence_number=r['mutation_pos']) #FIXME MAKE AA CHECK
if res.exists():
res = res.get()
else:
self.logger.error('Skipped due to no residue or mismatch AA ' +
r['protein'] + ' pos:' +
str(r['mutation_pos']) + ' AA:' +
r['mutation_from'])
# print('Skipped due to no residue or mismatch AA ' + r['protein'] + ' pos:'+str(r['mutation_pos']) + ' AA:'+r['mutation_from'],r['source_file'])
skipped += 1
continue
if r['ligand_class']:
try:
l_role, created = LigandRole.objects.get_or_create(
name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]
}) # FIXME this should not be needed
except Exception as e:
if LigandRole.objects.filter(
slug=slugify(r['ligand_class'])[:50]).exists():
l_role = LigandRole.objects.get(
slug=slugify(r['ligand_class'])[:50])
if l_role.name == slugify(r['ligand_class'])[:50]:
#if name of role is same as slug, then it was created by constructs script, replace it
l_role.name = r['ligand_class']
l_role.save()
else:
print(e)
print("Error with", r['ligand_class'],
slugify(r['ligand_class'])[:50])
l_role, created = LigandRole.objects.get_or_create(
slug=slugify(r['ligand_class'])
[:50]) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(
type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(
func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r['exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(
qual=r['exp_mu_effect_qual'],
prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
# if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r['opt_sign'] or r['opt_percentage'] or r['opt_qual'] or r['opt_agonist']:
# exp_opt_id, created = MutationOptional.objects.get_or_create(type=r['opt_type'], wt=r['opt_wt'], mu=r['opt_mu'], sign=r['opt_sign'], percentage=r['opt_percentage'], qual=r['opt_qual'], agonist=r['opt_agonist'])
# else:
# exp_opt_id = None
try:
mutation, created = Mutation.objects.get_or_create(
amino_acid=r['mutation_to'], protein=protein, residue=res)
except IntegrityError:
mutation = Mutation.objects.get(amino_acid=r['mutation_to'],
protein=protein,
residue=res)
logtypes = ['pEC50', 'pIC50', 'pK']
foldchange = 0
typefold = ''
if r['exp_wt_value'] != 0 and r[
'exp_mu_value_raw'] != 0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")",
r['exp_type']): #-log values!
try:
foldchange = round(
math.pow(10, -r['exp_mu_value_raw']) /
pow(10, -r['exp_wt_value']), 3)
except:
print(r)
typefold = r['exp_type'] + "_log"
elif "%" == r['exp_wt_unit']:
# if % then it's a difference case, then lower value is bad. Otherwise it's conc and lower is better
foldchange = round(
r['exp_wt_value'] / r['exp_mu_value_raw'], 3)
else:
foldchange = round(
r['exp_mu_value_raw'] / r['exp_wt_value'], 3)
typefold = r['exp_type'] + "_not_log"
if foldchange > 0 and foldchange < 1 and foldchange != 0:
foldchange = -round((1 / foldchange), 3)
elif r['fold_effect'] != 0:
foldchange = round(r['fold_effect'], 3)
if foldchange < 1: foldchange = -round((1 / foldchange), 3)
r['fold_effect'] = foldchange
raw_experiment = self.insert_raw(r)
# raw_experiment.save()
bulk = MutationExperiment(
refs=pub,
review=pub_review,
submitting_group=r['submitting_group'],
data_container=r['data_container'],
data_container_number=r['data_container_number'],
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref=l_ref,
# raw = raw_experiment, #raw_experiment, OR None
# optional = exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual=exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value=r['exp_mu_value_raw'],
mu_sign=r['exp_mu_effect_sign'],
foldchange=foldchange,
opt_receptor_expression=r['opt_receptor_expression'],
opt_basal_activity=r['opt_basal_activity'],
opt_gain_of_activity=r['opt_gain_of_activity'],
opt_ligand_emax=r['opt_ligand_emax'],
opt_agonist=r['opt_agonist'],
)
# for line,val in r.items():
# val = str(val)
# if len(val)>100:
# print(line,"too long",val)
# mut_id = obj.id
bulk_r.append(raw_experiment)
bulk_m.append(bulk)
# try:
# bulk.save()
# except Exception as e:
# print(e)
# print(r)
# break
#print('saved ',r['source_file'])
inserted += 1
end = time.time()
diff = round(end - current, 2)
#print(diff)
self.logger.info('Parsed ' + str(c) +
' mutant data entries. Skipped ' + str(skipped))
current = time.time()
raws = MutationRaw.objects.bulk_create(bulk_r)
for i, me in enumerate(bulk_m):
me.raw = raws[i]
MutationExperiment.objects.bulk_create(bulk_m)
end = time.time()
diff = round(end - current, 2)
# current_sheet
diff_2 = round(end - current_sheet, 2)
print("overall", diff_2, "bulk", diff, len(bulk_m), "skipped",
str(skipped))
sorted_missing_proteins = sorted(missing_proteins.items(),
key=operator.itemgetter(1),
reverse=True)
def get_or_make_ligand(ligand_id,type_id, name = None):
if type_id=='PubChem CID' or type_id=='SMILES':
if type_id=='PubChem CID':
pubchem_lookup_value = 'cid'
elif type_id=='SMILES':
pubchem_lookup_value = 'smiles'
try:
web_resource = WebResource.objects.get(slug='pubchem')
except:
# abort if pdb resource is not found
raise Exception('PubChem resource not found, aborting!')
if name:
ligand_name = name
else:
ligand_name = False
try:
# if this name is canonical and it has a ligand record already
if (ligand_name==False):
l = None
ls = Ligand.objects.filter(canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
for ligand in ls:
l = ligand
#print (l)
break
if l == None:
l = Ligand.objects.get(canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
else:
l = Ligand.objects.get(name=ligand_name, canonical=True,
properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id)
#l = Ligand.objects.get(name=ligand_name, canonical=True,
# properities__web_links__web_resource=web_resource,
# properities__web_links__index=ligand_id)
#
except Ligand.DoesNotExist:
try:
# if exists under different name
l_canonical = Ligand.objects.get(properities__web_links__web_resource=web_resource,
properities__web_links__index=ligand_id, canonical=True)
#print (created)
try:
l, created = Ligand.objects.get_or_create(properities = l_canonical.properities,
name = ligand_name, canonical = False)
except IntegrityError:
l = Ligand.objects.get(properities = l_canonical.properities,
name = ligand_name, canonical = False)
except Ligand.DoesNotExist:
# fetch ligand from pubchem
default_ligand_type = 'Small molecule'
lt, created = LigandType.objects.get_or_create(slug=slugify(default_ligand_type),
defaults={'name': default_ligand_type})
l = Ligand()
#print (ligand_name)
l = l.load_from_pubchem(pubchem_lookup_value, ligand_id, lt, ligand_name)
#print (l)
if l == None and type_id=='SMILES': #insert manually if smiles and unfound in pubchem
try:
l = Ligand.objects.get(name=ligand_name, canonical=True,
properities__smiles=ligand_id)
except Ligand.DoesNotExist:
try:
l = Ligand.objects.get(name__startswith=ligand_name, canonical=True,properities__smiles=ligand_id) #if no properities exist
except Ligand.DoesNotExist:
try:
l = Ligand.objects.get(name=ligand_name, canonical=True,properities__smiles=None) #if no properities exist
l.properities.smiles = ligand_id
l.properities.save()
l.save()
except Ligand.DoesNotExist:
## now insert a new ligand, but first make sure name is unique
if Ligand.objects.filter(name=ligand_name).exists():
ls = Ligand.objects.filter(name__startswith=ligand_name, canonical=True).order_by("pk")
for l_temp in ls:
last = l_temp.name.split("_")[-1]
if last==ligand_name: #no addition yet
ligand_name = ligand_name +"_1"
else:
ligand_name = ligand_name +"_"+str(int(last)+1)
l = Ligand()
l.name = ligand_name
lp = LigandProperities()
lp.smiles = ligand_id
lp.ligand_type = lt
lp.save()
l.properities = lp
l.canonical = True #maybe false, but that would break stuff.
l.ambigious_alias = False
try:
l.save()
except IntegrityError:
l = Ligand.objects.get(name=ligand_name, canonical=True)
elif name:
# if this name is canonical and it has a ligand record already
if Ligand.objects.filter(name=name, canonical=True).exists():
l = Ligand.objects.get(name=name, canonical=True)
# if this matches an alias that only has "one" parent canonical name - eg distinct
elif Ligand.objects.filter(name=name, canonical=False,
ambigious_alias=False).exists():
l = Ligand.objects.get(name=name, canonical=False, ambigious_alias=False)
# if this matches an alias that only has several canonical parents, must investigate, start
# with empty.
elif Ligand.objects.filter(name=name, canonical=False,
ambigious_alias=True).exists():
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = name
l.canonical = False
l.ambigious_alias = True
l.save()
l.load_by_name(name)
# if neither a canonical or alias exists, create the records. Remember to check for
# canonical / alias status.
else:
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = str(name)
l.canonical = True
l.ambigious_alias = False
try:
l.save()
l.load_by_name(str(name))
except IntegrityError:
l = Ligand.objects.get(name=str(name), canonical=True)
else:
l = None
return l
def build_ligand_properties(self):
lp = LigandProperities()
lt = LigandType.objects.get(name = 'small molecule')
lp.ligand_type = lt
lp.smiles = None
lp.inchikey = None
lp.sequence= None
lp.mw = None
lp.rotatable_bonds = None
lp.hacc = None
lp.hdon = None
lp.logp = None
lp.save()
self.logger.info("Could not create ligand, empty is returned")
return lp
def main_func(self, positions, iteration, count, lock):
# print(positions,iteration,count,lock)
ligands = self.ligand_dump
while count.value < len(ligands):
with lock:
l = ligands[count.value]
count.value += 1
if count.value % 10000 == 0:
print('{} Status {} out of {}'.format(
datetime.datetime.strftime(datetime.datetime.now(),
'%Y-%m-%d %H:%M:%S'),
count.value, len(ligands)))
if 'logp' not in l:
# temp skip to only use "full" annotated ligands
continue
lp = LigandProperities.objects.filter(
inchikey=l['inchikey']).first()
ligand = None
if lp:
# Check if inchikey is there
ligand = Ligand.objects.filter(
name=l['name'], properities=lp).prefetch_related(
'properities__ligand_type', 'properities__web_links',
'properities__vendors').first()
# The name with corresponding inchikey is there, assume all is good and skip.
# Will add links to make sure they're there.
if not ligand:
if lp:
print(l['name'], 'is there! (but not by name, only inchi')
ligand = Ligand()
ligand.properities = lp
ligand.name = l['name']
ligand.canonical = l['canonical']
ligand.ambigious_alias = l['ambigious_alias']
ligand.save()
else:
# No ligand seems to match by inchikey -- start creating it.
# Make LigandProperities first
lt, created = LigandType.objects.get_or_create(
slug=l['ligand_type__slug'],
defaults={'name': l['ligand_type__name']})
lp = LigandProperities()
lp.inchikey = l['inchikey']
lp.smiles = l['smiles']
lp.mw = l['mw']
lp.logp = l['logp']
lp.rotatable_bonds = l['rotatable_bonds']
lp.hacc = l['hacc']
lp.hdon = l['hdon']
lp.ligand_type = lt
lp.save()
ligand = Ligand()
ligand.properities = lp
ligand.name = l['name']
ligand.canonical = l['canonical']
ligand.ambigious_alias = l['ambigious_alias']
ligand.save()
# create links - impossible to make duplicates so no need to check if there already
if ligand.properities.web_links.count() < len(l['web_links']):
for link in l['web_links']:
wr = WebResource.objects.get(slug=link['web_resource'])
wl, created = WebLink.objects.get_or_create(
index=link['index'], web_resource=wr)
ligand.properities.web_links.add(wl)
# create vendors - impossible to make duplicates so no need to check if there already
if ligand.properities.vendors.count() < len(l['vendors']):
for link in l['vendors']:
lv = LigandVendors.objects.get(slug=link['vendor_slug'])
check = LigandVendorLink.objects.filter(
sid=link['sid']).exists()
if not check:
lvl = LigandVendorLink()
lvl.sid = link['sid']
lvl.vendor = lv
lvl.lp = ligand.properities
lvl.vendor_external_id = link['vendor_external_id']
lvl.url = link['url']
lvl.save()
def create_mutant_data(self, filenames):
self.logger.info('CREATING MUTANT DATA')
# what files should be parsed?
if not filenames:
filenames = os.listdir(self.structure_data_dir)
missing_proteins = {}
mutants_for_proteins = {}
for source_file in filenames:
source_file_path = os.sep.join(
[self.structure_data_dir, source_file])
if os.path.isfile(source_file_path) and source_file[0] != '.':
self.logger.info('Reading file {}'.format(source_file_path))
# read the yaml file
if source_file[-4:] == 'xlsx' or source_file[-3:] == 'xls':
rows = self.loaddatafromexcel(source_file_path)
rows = self.analyse_rows(rows)
elif source_file[-4:] == 'yaml':
rows = yaml.load(open(source_file_path, 'r'))
temp = []
for r in rows:
d = {}
d['reference'] = r['pubmed']
d['protein'] = r['entry_name'].replace("__",
"_").lower()
d['mutation_pos'] = r['seq']
d['mutation_from'] = r['from_res']
d['mutation_to'] = r['to_res']
d['ligand_name'] = ''
d['ligand_type'] = ''
d['ligand_id'] = ''
d['ligand_class'] = ''
d['exp_type'] = ''
d['exp_func'] = ''
d['exp_wt_value'] = 0
d['exp_wt_unit'] = ''
d['exp_mu_effect_sign'] = ''
d['exp_mu_value_raw'] = 0
d['fold_effect'] = 0
d['exp_mu_effect_qual'] = ''
d['exp_mu_effect_ligand_prop'] = ''
d['exp_mu_ligand_ref'] = ''
d['opt_type'] = ''
d['opt_wt'] = 0
d['opt_mu'] = 0
d['opt_sign'] = ''
d['opt_percentage'] = 0
d['opt_qual'] = ''
d['opt_agonist'] = ''
if len(d['mutation_to']) > 1 or len(
d['mutation_from']
) > 1: #if something is off with amino acid
continue
temp.append(d)
rows = temp
else:
self.logger.info('unknown format'.source_file)
continue
c = 0
skipped = 0
inserted = 0
for r in rows:
c += 1
if c % 1000 == 0:
self.logger.info('Parsed ' + str(c) +
' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
try:
pub = Publication.objects.get(
web_link__index=r['reference'],
web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = WebLink.objects.get(
index=r['reference'],
web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(
index=r['reference'],
web_resource=WebResource.objects.get(
slug=pub_type))
pub.web_link = wl
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' +
str(r['reference']) + ' ' +
pub_type)
continue #if something off with publication, skip.
if r['ligand_type'] == 'PubChem CID' or r[
'ligand_type'] == 'SMILES':
if r['ligand_type'] == 'PubChem CID':
pubchem_lookup_value = 'cid'
elif r['ligand_type'] == 'SMILES':
pubchem_lookup_value = 'smiles'
try:
web_resource = WebResource.objects.get(
slug='pubchem')
except:
# abort if pdb resource is not found
raise Exception(
'PubChem resource not found, aborting!')
if 'ligand_name' in r and r['ligand_name']:
ligand_name = str(r['ligand_name'])
else:
ligand_name = False
try:
# if this name is canonical and it has a ligand record already
l = Ligand.objects.get(
name=ligand_name,
canonical=True,
properities__web_links__web_resource=
web_resource,
properities__web_links__index=r['ligand_id'])
except Ligand.DoesNotExist:
try:
# if exists under different name
l_canonical = Ligand.objects.get(
properities__web_links__web_resource=
web_resource,
properities__web_links__index=r[
'ligand_id'],
canonical=True)
l, created = Ligand.objects.get_or_create(
properities=l_canonical.properities,
name=ligand_name,
canonical=False)
if created:
self.logger.info(
'Created ligand {}'.format(l.name))
except Ligand.DoesNotExist:
# fetch ligand from pubchem
default_ligand_type = 'Small molecule'
lt, created = LigandType.objects.get_or_create(
slug=slugify(default_ligand_type),
defaults={'name': default_ligand_type})
l = Ligand()
l = l.load_from_pubchem(
pubchem_lookup_value, r['ligand_id'], lt,
ligand_name)
elif r['ligand_name']:
# if this name is canonical and it has a ligand record already
if Ligand.objects.filter(name=r['ligand_name'],
canonical=True).exists():
l = Ligand.objects.get(name=r['ligand_name'],
canonical=True)
# if this matches an alias that only has "one" parent canonical name - eg distinct
elif Ligand.objects.filter(
name=r['ligand_name'],
canonical=False,
ambigious_alias=False).exists():
l = Ligand.objects.get(name=r['ligand_name'],
canonical=False,
ambigious_alias=False)
# if this matches an alias that only has several canonical parents, must investigate, start
# with empty.
elif Ligand.objects.filter(
name=r['ligand_name'],
canonical=False,
ambigious_alias=True).exists():
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = r['ligand_name']
l.canonical = False
l.ambigious_alias = True
l.save()
l.load_by_name(r['ligand_name'])
# if neither a canonical or alias exists, create the records. Remember to check for
# canonical / alias status.
else:
lp = LigandProperities()
lp.save()
l = Ligand()
l.properities = lp
l.name = str(r['ligand_name'])
l.canonical = True
l.ambigious_alias = False
l.save()
l.load_by_name(str(r['ligand_name']))
else:
l = None
if Ligand.objects.filter(
name=r['exp_mu_ligand_ref'], canonical=True
).exists(
): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=True)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False
).exists(
): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=False)
elif Ligand.objects.filter(
name=r['exp_mu_ligand_ref'],
canonical=False,
ambigious_alias=True
).exists(
): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
else:
l_ref = None
protein_id = 0
residue_id = 0
protein = Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein = protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
self.logger.error('Skipped due to no protein ' +
r['protein'])
continue
res = Residue.objects.filter(
protein_conformation__protein=protein,
sequence_number=r['mutation_pos'])
if res.exists():
res = res.get()
else:
self.logger.error('Skipped due to no residue ' +
r['protein'] + ' pos:' +
str(r['mutation_pos']))
skipped += 1
continue
if r['ligand_class']:
l_role, created = LigandRole.objects.get_or_create(
name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]
}) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(
type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(
func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r[
'exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(
qual=r['exp_mu_effect_qual'],
prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r[
'opt_sign'] or r['opt_percentage'] or r[
'opt_qual'] or r['opt_agonist']:
exp_opt_id, created = MutationOptional.objects.get_or_create(
type=r['opt_type'],
wt=r['opt_wt'],
mu=r['opt_mu'],
sign=r['opt_sign'],
percentage=r['opt_percentage'],
qual=r['opt_qual'],
agonist=r['opt_agonist'])
else:
exp_opt_id = None
mutation, created = Mutation.objects.get_or_create(
amino_acid=r['mutation_to'],
protein=protein,
residue=res)
logtypes = ['pEC50', 'pIC50', 'pK']
foldchange = 0
typefold = ''
if r['exp_wt_value'] != 0 and r[
'exp_mu_value_raw'] != 0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")",
r['exp_type']): #-log values!
foldchange = round(
math.pow(10, -r['exp_mu_value_raw']) /
pow(10, -r['exp_wt_value']), 3)
typefold = r['exp_type'] + "_log"
else:
foldchange = round(
r['exp_mu_value_raw'] / r['exp_wt_value'], 3)
typefold = r['exp_type'] + "_not_log"
if foldchange < 1 and foldchange != 0:
foldchange = -round((1 / foldchange), 3)
elif r['fold_effect'] != 0:
foldchange = round(r['fold_effect'], 3)
if foldchange < 1:
foldchange = -round((1 / foldchange), 3)
raw_experiment = self.insert_raw(r)
obj, created = MutationExperiment.objects.get_or_create(
refs=pub,
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref=l_ref,
raw=raw_experiment,
optional=exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual=exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value=r['exp_mu_value_raw'],
mu_sign=r['exp_mu_effect_sign'],
foldchange=foldchange)
mut_id = obj.id
inserted += 1
self.logger.info('Parsed ' + str(c) +
' mutant data entries. Skipped ' +
str(skipped))
sorted_missing_proteins = sorted(missing_proteins.items(),
key=operator.itemgetter(1),
reverse=True)
sorted_mutants_for_proteins = sorted(mutants_for_proteins.items(),
key=operator.itemgetter(1),
reverse=True)
self.logger.info('COMPLETED CREATING MUTANTS')
def main_func(self, positions, iteration,count,lock):
# print(positions,iteration,count,lock)
ligands = self.ligand_dump
while count.value<len(ligands):
with lock:
l = ligands[count.value]
count.value +=1
if count.value % 10000 == 0:
print('{} Status {} out of {}'.format(
datetime.datetime.strftime(datetime.datetime.now(), '%Y-%m-%d %H:%M:%S'), count.value, len(ligands)))
if 'logp' not in l:
# temp skip to only use "full" annotated ligands
continue
lp = LigandProperities.objects.filter(inchikey=l['inchikey']).first()
ligand = None
if lp:
# Check if inchikey is there
ligand = Ligand.objects.filter(name=l['name'], properities=lp).prefetch_related('properities__ligand_type','properities__web_links','properities__vendors').first()
# The name with corresponding inchikey is there, assume all is good and skip.
# Will add links to make sure they're there.
if not ligand:
if lp:
print(l['name'],'is there! (but not by name, only inchi')
ligand = Ligand()
ligand.properities = lp
ligand.name = l['name']
ligand.canonical = l['canonical']
ligand.ambigious_alias = l['ambigious_alias']
ligand.save()
else:
# No ligand seems to match by inchikey -- start creating it.
# Make LigandProperities first
lt, created = LigandType.objects.get_or_create(slug=l['ligand_type__slug'],defaults = {'name':l['ligand_type__name']})
lp = LigandProperities()
lp.inchikey = l['inchikey']
lp.smiles = l['smiles']
lp.mw = l['mw']
lp.logp = l['logp']
lp.rotatable_bonds = l['rotatable_bonds']
lp.hacc = l['hacc']
lp.hdon = l['hdon']
lp.ligand_type = lt
lp.save()
ligand = Ligand()
ligand.properities = lp
ligand.name = l['name']
ligand.canonical = l['canonical']
ligand.ambigious_alias = l['ambigious_alias']
ligand.save()
# create links - impossible to make duplicates so no need to check if there already
if ligand.properities.web_links.count()<len(l['web_links']):
for link in l['web_links']:
wr = WebResource.objects.get(slug=link['web_resource'])
wl, created = WebLink.objects.get_or_create(index=link['index'], web_resource=wr)
ligand.properities.web_links.add(wl)
# create vendors - impossible to make duplicates so no need to check if there already
if ligand.properities.vendors.count()<len(l['vendors']):
for link in l['vendors']:
lv = LigandVendors.objects.get(slug = link['vendor_slug'])
check = LigandVendorLink.objects.filter(sid=link['sid']).exists()
if not check:
lvl = LigandVendorLink()
lvl.sid = link['sid']
lvl.vendor = lv
lvl.lp = ligand.properities
lvl.vendor_external_id = link['vendor_external_id']
lvl.url = link['url']
lvl.save()
def main_func(self, positions, iteration):
# filenames
if not positions[1]:
rows = self.data[positions[0]:]
else:
rows = self.data[positions[0]:positions[1]]
missing_proteins = {}
mutants_for_proteins = {}
c = 0
skipped = 0
inserted = 0
bulk_m = []
bulk_r = []
current_sheet = time.time()
for r in rows:
# print(source_file,c)
# PRINT IF ERRORS OCCUR
# self.logger.info('File '+str(r['source_file'])+' number '+str(c))
current = time.time()
c += 1
# if c%100==0:
# self.logger.info('Parsed '+str(c)+' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
float(r['review'])
r['review'] = str(int(r['review']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
if r['reference'] not in self.publication_cache:
try:
pub = Publication.objects.get(web_link__index=r['reference'], web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = WebLink.objects.get(index=r['reference'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(index=r['reference'],
web_resource = WebResource.objects.get(slug=pub_type))
pub.web_link = wl
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' + str(r['reference']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['reference']] = pub
else:
pub = self.publication_cache[r['reference']]
# print(r['review'],r['reference'])
if r['review'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
# print(r['review'],pub_type)
if r['review']:
if r['review'] not in self.publication_cache:
try:
pub_review = Publication.objects.get(web_link__index=r['review'], web_link__web_resource__slug=pub_type)
except Publication.DoesNotExist:
pub_review = Publication()
try:
pub_review.web_link = WebLink.objects.get(index=r['review'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
wl = WebLink.objects.create(index=r['review'],
web_resource = WebResource.objects.get(slug=pub_type))
pub_review.web_link = wl
if pub_type == 'doi':
pub_review.update_from_doi(doi=r['review'])
elif pub_type == 'pubmed':
pub_review.update_from_pubmed_data(index=r['review'])
try:
pub_review.save()
except:
self.logger.error('error with review ' + str(r['review']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['review']] = pub_review
else:
pub_review = self.publication_cache[r['review']]
else:
pub_review = None
l = None
if str(r['ligand_name']) in self.ligand_cache:
if r['ligand_id'] in self.ligand_cache[str(r['ligand_name'])]:
l = self.ligand_cache[str(r['ligand_name'])][r['ligand_id']]
else:
self.ligand_cache[str(r['ligand_name'])] = {}
if not l:
l = get_or_make_ligand(r['ligand_id'],r['ligand_type'],str(r['ligand_name']))
self.ligand_cache[str(r['ligand_name'])][r['ligand_id']] = l
l_ref = None
if str(r['exp_mu_ligand_ref']) in self.ref_ligand_cache:
l_ref = self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])]
else:
if Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=True).exists(): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=True)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False).exists(): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=True).exists(): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False).exists(): #amigious_alias not specified
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False)
l_ref.ambigious_alias = False
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
try:
l_ref.save()
except IntegrityError:
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False)
print("error failing ligand, duplicate?")
# logger.error("FAILED SAVING LIGAND, duplicate?")
else:
l_ref = None
self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])] = l_ref
protein_id = 0
residue_id = 0
protein=Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein=protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
self.logger.error('Skipped due to no protein '+ r['protein'])
continue
res=Residue.objects.filter(protein_conformation__protein=protein,amino_acid=r['mutation_from'],sequence_number=r['mutation_pos']) #FIXME MAKE AA CHECK
if res.exists():
res=res.get()
else:
self.logger.error('Skipped due to no residue or mismatch AA ' + r['protein'] + ' pos:'+str(r['mutation_pos']) + ' AA:'+r['mutation_from'])
skipped += 1
continue
if r['ligand_class']:
l_role, created = LigandRole.objects.get_or_create(name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]}) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r['exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(qual=r['exp_mu_effect_qual'], prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r['opt_sign'] or r['opt_percentage'] or r['opt_qual'] or r['opt_agonist']:
exp_opt_id, created = MutationOptional.objects.get_or_create(type=r['opt_type'], wt=r['opt_wt'], mu=r['opt_mu'], sign=r['opt_sign'], percentage=r['opt_percentage'], qual=r['opt_qual'], agonist=r['opt_agonist'])
else:
exp_opt_id = None
try:
mutation, created = Mutation.objects.get_or_create(amino_acid=r['mutation_to'],protein=protein, residue=res)
except IntegrityError:
mutation = Mutation.objects.get(amino_acid=r['mutation_to'],protein=protein, residue=res)
logtypes = ['pEC50','pIC50','pK']
foldchange = 0
typefold = ''
if r['exp_wt_value']!=0 and r['exp_mu_value_raw']!=0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")", r['exp_type']): #-log values!
foldchange = round(math.pow(10,-r['exp_mu_value_raw'])/pow(10,-r['exp_wt_value']),3);
typefold = r['exp_type']+"_log"
else:
foldchange = round(r['exp_mu_value_raw']/r['exp_wt_value'],3);
typefold = r['exp_type']+"_not_log"
if foldchange<1 and foldchange!=0:
foldchange = -round((1/foldchange),3)
elif r['fold_effect']!=0:
foldchange = round(r['fold_effect'],3);
if foldchange<1: foldchange = -round((1/foldchange),3);
raw_experiment = self.insert_raw(r)
bulk = MutationExperiment(
refs=pub,
review=pub_review,
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref = l_ref,
#raw = raw_experiment, #raw_experiment, OR None
optional = exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual = exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value = r['exp_mu_value_raw'],
mu_sign = r['exp_mu_effect_sign'],
foldchange = foldchange
)
# mut_id = obj.id
bulk_r.append(raw_experiment)
bulk_m.append(bulk)
inserted += 1
end = time.time()
diff = round(end - current,2)
#print(diff)
self.logger.info('Parsed '+str(c)+' mutant data entries. Skipped '+str(skipped))
current = time.time()
raws = MutationRaw.objects.bulk_create(bulk_r)
for i,me in enumerate(bulk_m):
me.raw = raws[i]
MutationExperiment.objects.bulk_create(bulk_m)
end = time.time()
diff = round(end - current,2)
current_sheet
diff_2 = round(end - current_sheet,2)
print("overall",diff_2,"bulk",diff,len(bulk_m),"skipped",str(skipped))
def main_func(self, positions, iteration,count,lock):
# filenames
# if not positions[1]:
# rows = self.data[positions[0]:]
# else:
# rows = self.data[positions[0]:positions[1]]
missing_proteins = {}
mutants_for_proteins = {}
wrong_uniport_ids = {}
c = 0
skipped = 0
inserted = 0
bulk_m = []
bulk_r = []
current_sheet = time.time()
rows = self.data_all
while count.value<len(rows):
with lock:
r = rows[count.value]
count.value +=1
# for r in rows:
# print(r['source_file'],c)
# PRINT IF ERRORS OCCUR
#self.logger.info('File '+str(r['source_file'])+' number '+str(c))
current = time.time()
c += 1
# if c%100==0:
# self.logger.info('Parsed '+str(c)+' mutant data entries')
# publication
try: #fix if it thinks it's float.
float(r['reference'])
r['reference'] = str(int(r['reference']))
float(r['review'])
r['review'] = str(int(r['review']))
except ValueError:
pass
if r['reference'].isdigit(): #assume pubmed
pub_type = 'pubmed'
else: #assume doi
pub_type = 'doi'
if r['reference'] not in self.publication_cache:
try:
wl = WebLink.objects.get(index=r['reference'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
try:
wl = WebLink.objects.create(index=r['reference'],
web_resource = WebResource.objects.get(slug=pub_type))
except IntegrityError:
wl = WebLink.objects.get(index=r['reference'], web_resource__slug=pub_type)
try:
pub = Publication.objects.get(web_link=wl)
except Publication.DoesNotExist:
pub = Publication()
try:
pub.web_link = wl
pub.save()
except IntegrityError:
pub = Publication.objects.get(web_link=wl)
if pub_type == 'doi':
pub.update_from_doi(doi=r['reference'])
elif pub_type == 'pubmed':
pub.update_from_pubmed_data(index=r['reference'])
try:
pub.save()
except:
self.logger.error('error with reference ' + str(r['reference']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['reference']] = pub
else:
pub = self.publication_cache[r['reference']]
# print(r['review'],r['reference'])
if r['review'].isdigit(): #assume pubmed
pub_type = 'pubmed'
elif r['review'].startswith('http'):
pub_type = 'raw_link'
else: #assume doi
pub_type = 'doi'
# print(r['review'],pub_type)
if r['review']:
if r['review'] not in self.publication_cache:
try:
wl = WebLink.objects.get(index=r['review'], web_resource__slug=pub_type)
except WebLink.DoesNotExist:
try:
wl = WebLink.objects.create(index=r['review'],
web_resource = WebResource.objects.get(slug=pub_type))
except IntegrityError:
wl = WebLink.objects.get(index=r['review'], web_resource__slug=pub_type)
try:
pub_review = Publication.objects.get(web_link=wl)
except Publication.DoesNotExist:
pub_review = Publication()
try:
pub_review.web_link = wl
pub_review.save()
except IntegrityError:
pub_review = Publication.objects.get(web_link=wl)
if pub_type == 'doi':
pub_review.update_from_doi(doi=r['review'])
elif pub_type == 'pubmed':
pub_review.update_from_pubmed_data(index=r['review'])
try:
pub_review.save()
except:
self.logger.error('error with review ' + str(r['review']) + ' ' + pub_type)
continue #if something off with publication, skip.
self.publication_cache[r['review']] = pub_review
else:
pub_review = self.publication_cache[r['review']]
else:
pub_review = None
l = None
if str(r['ligand_name']) in self.ligand_cache:
if r['ligand_id'] in self.ligand_cache[str(r['ligand_name'])]:
l = self.ligand_cache[str(r['ligand_name'])][r['ligand_id']]
else:
self.ligand_cache[str(r['ligand_name'])] = {}
if not l:
try:
l = get_or_make_ligand(r['ligand_id'],r['ligand_type'],str(r['ligand_name']))
except Exception as msg:
print('Something errored with ligand, aborting entry of mutation',r['ligand_name'],r['ligand_type'],r['ligand_id'],r['source_file'])
print(msg)
traceback.print_exc()
continue
self.ligand_cache[str(r['ligand_name'])][r['ligand_id']] = l
l_ref = None
if str(r['exp_mu_ligand_ref']) in self.ref_ligand_cache:
l_ref = self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])]
else:
if Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=True).exists(): #if this name is canonical and it has a ligand record already
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=True)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False).exists(): #if this matches an alias that only has "one" parent canonical name - eg distinct
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=False)
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False, ambigious_alias=True).exists(): #if this matches an alias that only has several canonical parents, must investigate, start with empty.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = False
l_ref.ambigious_alias = True
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
l_ref.save()
elif Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=False).exists(): #amigious_alias not specified
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False)
l_ref.ambigious_alias = False
l_ref.save()
elif r['exp_mu_ligand_ref']: #if neither a canonical or alias exists, create the records. Remember to check for canonical / alias status.
lp = LigandProperities()
lp.save()
l_ref = Ligand()
l_ref.properities = lp
l_ref.name = r['exp_mu_ligand_ref']
l_ref.canonical = True
l_ref.ambigious_alias = False
try:
l_ref.save()
l_ref.load_by_name(r['exp_mu_ligand_ref'])
except IntegrityError:
if Ligand.objects.filter(name=r['exp_mu_ligand_ref'], canonical=True).exists():
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=True)
else:
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False)
# print("error failing ligand, duplicate?")
try:
l_ref.save()
except IntegrityError:
l_ref = Ligand.objects.get(name=r['exp_mu_ligand_ref'], canonical=False)
# print("error failing ligand, duplicate?")
# logger.error("FAILED SAVING LIGAND, duplicate?")
else:
l_ref = None
self.ref_ligand_cache[str(r['exp_mu_ligand_ref'])] = l_ref
protein_id = 0
residue_id = 0
protein=Protein.objects.filter(entry_name=r['protein'])
if protein.exists():
protein=protein.get()
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
elif r['protein'] not in missing_proteins:
try:
r['protein'] = wrong_uniport_ids[r['protein']]
real_uniprot = wrong_uniport_ids[r['protein']]
protein=Protein.objects.get(entry_name=r['protein'])
# print('fetched with lookup table',r['protein'])
except:
# look for it as uniprot
protein=Protein.objects.filter(web_links__web_resource__slug='uniprot', web_links__index=r['protein'].upper())
if protein.exists():
protein=protein.get()
real_uniprot = protein.entry_name
if r['protein'] in mutants_for_proteins:
mutants_for_proteins[r['protein']] += 1
else:
mutants_for_proteins[r['protein']] = 1
else:
# Try to lookup in uniprot to catch typing errors / variants in entry_name
url = 'http://www.uniprot.org/uniprot/$index.xml'
cache_dir = ['uniprot', 'id']
uniprot_protein = fetch_from_web_api(url, r['protein'], cache_dir, xml = True)
try:
real_uniprot = uniprot_protein.find('.//{http://uniprot.org/uniprot}name').text.lower()
protein=Protein.objects.get(entry_name=real_uniprot)
except:
skipped += 1
if r['protein'] in missing_proteins:
missing_proteins[r['protein']] += 1
else:
missing_proteins[r['protein']] = 1
# print('Skipped due to no protein '+ r['protein'])
self.logger.error('Skipped due to no protein '+ r['protein'])
continue
wrong_uniport_ids[r['protein']] = protein.entry_name
r['protein'] = real_uniprot
else:
missing_proteins[r['protein']] += 1
continue
res=Residue.objects.filter(protein_conformation__protein=protein,amino_acid=r['mutation_from'],sequence_number=r['mutation_pos']) #FIXME MAKE AA CHECK
if res.exists():
res=res.get()
else:
self.logger.error('Skipped due to no residue or mismatch AA ' + r['protein'] + ' pos:'+str(r['mutation_pos']) + ' AA:'+r['mutation_from'])
# print('Skipped due to no residue or mismatch AA ' + r['protein'] + ' pos:'+str(r['mutation_pos']) + ' AA:'+r['mutation_from'],r['source_file'])
skipped += 1
continue
if r['ligand_class']:
try:
l_role, created = LigandRole.objects.get_or_create(name=r['ligand_class'],
defaults={'slug': slugify(r['ligand_class'])[:50]}) # FIXME this should not be needed
except Exception as e:
if LigandRole.objects.filter(slug=slugify(r['ligand_class'])[:50]).exists():
l_role = LigandRole.objects.get(slug=slugify(r['ligand_class'])[:50])
if l_role.name == slugify(r['ligand_class'])[:50]:
#if name of role is same as slug, then it was created by constructs script, replace it
l_role.name = r['ligand_class']
l_role.save()
else:
print(e)
print("Error with",r['ligand_class'],slugify(r['ligand_class'])[:50] )
l_role, created = LigandRole.objects.get_or_create(slug=slugify(r['ligand_class'])[:50]) # FIXME this should not be needed
else:
l_role = None
if r['exp_type']:
exp_type_id, created = MutationExperimentalType.objects.get_or_create(type=r['exp_type'])
else:
exp_type_id = None
if r['exp_func']:
exp_func_id, created = MutationFunc.objects.get_or_create(func=r['exp_func'])
else:
exp_func_id = None
if r['exp_mu_effect_ligand_prop'] or r['exp_mu_effect_qual']:
exp_qual_id, created = MutationQual.objects.get_or_create(qual=r['exp_mu_effect_qual'], prop=r['exp_mu_effect_ligand_prop'])
else:
exp_qual_id = None
# if r['opt_type'] or r['opt_wt'] or r['opt_mu'] or r['opt_sign'] or r['opt_percentage'] or r['opt_qual'] or r['opt_agonist']:
# exp_opt_id, created = MutationOptional.objects.get_or_create(type=r['opt_type'], wt=r['opt_wt'], mu=r['opt_mu'], sign=r['opt_sign'], percentage=r['opt_percentage'], qual=r['opt_qual'], agonist=r['opt_agonist'])
# else:
# exp_opt_id = None
try:
mutation, created = Mutation.objects.get_or_create(amino_acid=r['mutation_to'],protein=protein, residue=res)
except IntegrityError:
mutation = Mutation.objects.get(amino_acid=r['mutation_to'],protein=protein, residue=res)
logtypes = ['pEC50','pIC50','pK']
foldchange = 0
typefold = ''
if r['exp_wt_value']!=0 and r['exp_mu_value_raw']!=0: #fix for new format
if re.match("(" + ")|(".join(logtypes) + ")", r['exp_type']): #-log values!
try:
foldchange = round(math.pow(10,-r['exp_mu_value_raw'])/pow(10,-r['exp_wt_value']),3);
except:
print(r)
typefold = r['exp_type']+"_log"
elif "%"==r['exp_wt_unit']:
# if % then it's a difference case, then lower value is bad. Otherwise it's conc and lower is better
foldchange = round(r['exp_wt_value']/r['exp_mu_value_raw'],3);
else:
foldchange = round(r['exp_mu_value_raw']/r['exp_wt_value'],3);
typefold = r['exp_type']+"_not_log"
if foldchange>0 and foldchange<1 and foldchange!=0:
foldchange = -round((1/foldchange),3)
elif r['fold_effect']!=0:
foldchange = round(r['fold_effect'],3);
if foldchange<1: foldchange = -round((1/foldchange),3);
r['fold_effect'] = foldchange
raw_experiment = self.insert_raw(r)
# raw_experiment.save()
bulk = MutationExperiment(
refs=pub,
review=pub_review,
submitting_group = r['submitting_group'],
data_container = r['data_container'],
data_container_number = r['data_container_number'],
protein=protein,
residue=res,
ligand=l,
ligand_role=l_role,
ligand_ref = l_ref,
# raw = raw_experiment, #raw_experiment, OR None
# optional = exp_opt_id,
exp_type=exp_type_id,
exp_func=exp_func_id,
exp_qual = exp_qual_id,
mutation=mutation,
wt_value=r['exp_wt_value'], #
wt_unit=r['exp_wt_unit'],
mu_value = r['exp_mu_value_raw'],
mu_sign = r['exp_mu_effect_sign'],
foldchange = foldchange,
opt_receptor_expression = r['opt_receptor_expression'],
opt_basal_activity = r['opt_basal_activity'],
opt_gain_of_activity = r['opt_gain_of_activity'],
opt_ligand_emax = r['opt_ligand_emax'],
opt_agonist = r['opt_agonist'],
)
# for line,val in r.items():
# val = str(val)
# if len(val)>100:
# print(line,"too long",val)
# mut_id = obj.id
bulk_r.append(raw_experiment)
bulk_m.append(bulk)
# try:
# bulk.save()
# except Exception as e:
# print(e)
# print(r)
# break
#print('saved ',r['source_file'])
inserted += 1
end = time.time()
diff = round(end - current,2)
#print(diff)
self.logger.info('Parsed '+str(c)+' mutant data entries. Skipped '+str(skipped))
current = time.time()
raws = MutationRaw.objects.bulk_create(bulk_r)
for i,me in enumerate(bulk_m):
me.raw = raws[i]
MutationExperiment.objects.bulk_create(bulk_m)
end = time.time()
diff = round(end - current,2)
current_sheet
diff_2 = round(end - current_sheet,2)
print("overall",diff_2,"bulk",diff,len(bulk_m),"skipped",str(skipped))
sorted_missing_proteins = sorted(missing_proteins.items(), key=operator.itemgetter(1),reverse=True)
