def df_to_db(spec_df):
#added_by_dict = make_added_by_user_dict()
print('Saving biocatdb_2 excel to mongodb..')
for i, row in spec_df.iterrows():
html_doi = str(row['html_doi'])
doi = str(row['html_doi'])
added_by_string = str(row['added_by'])
list_html_to_remove = [
'https://doi.org/', 'http://doi.org/', 'http://dx.doi.org/'
]
for to_remove in list_html_to_remove:
if to_remove in doi:
doi = html_doi.replace(to_remove, '')
if len(Paper.objects(doi=doi)) == 0:
paper = Paper(short_citation=str(row['short_citation']),
html=html_doi,
doi=doi)
paper = paper.save()
print(f"{row['short_citation']} added")
else:
paper = Paper.objects(doi=doi)[0]
if row['enzyme_type'] is not None and row['enzyme_type'] != '' and type(
row['enzyme_type']) == str:
if len(EnzymeType.objects(enzyme_type=row['enzyme_type'])) == 0:
enz_type = EnzymeType(enzyme_type=row['enzyme_type'],
description='')
enz_type.save()
if row['enzyme_name'] is not None and row['enzyme_name'] != '' and type(
row['enzyme_name']) == str:
if len(Sequence.objects(enzyme_name=row['enzyme_name'])) == 0:
seq = Sequence(enzyme_name=check_is_nan(row['enzyme_name']),
enzyme_type=check_is_nan(row['enzyme_type']),
papers=[paper])
seq.save()
else:
seq = Sequence.objects(enzyme_name=row['enzyme_name'])[0]
if paper not in seq.papers:
seq.papers.append(paper)
seq = seq.save()
if row['binary'] == 1:
binary = True
else:
binary = False
if row['auto_generated'] == 1:
auto_gen = True
else:
auto_gen = False
activity = Activity(
enzyme_type=check_is_nan(row['enzyme_type']),
enzyme_name=check_is_nan(row['enzyme_name']),
reaction=check_is_nan(row['reaction']),
short_citation=check_is_nan(row['short_citation']),
html_doi=check_is_nan(row['html_doi']),
added_by_string=added_by_string,
paper=paper,
cascade_num=check_is_nan(row['cascade_num']),
substrate_1_smiles=get_smile(row['substrate_1_smiles']),
substrate_2_smiles=get_smile(row['substrate_2_smiles']),
product_1_smiles=get_smile(row['product_1_smiles']),
temperature=check_is_nan(row['temperature']),
ph=check_is_nan(row['ph']),
solvent=check_is_nan(row['solvent']),
other_conditions=check_is_nan(row['other_conditions']),
notes=check_is_nan(row['notes']),
reaction_vol=check_is_nan(row['reaction_vol']),
formulation=check_is_nan(row['formulation']),
biocat_conc=check_is_nan(row['biocat_conc']),
kcat=check_is_float(row['kcat']),
km=check_is_float(row['km']),
mw=check_is_float(row['mw']),
substrate_1_conc=check_is_nan(row['substrate_1_conc']),
substrate_2_conc=check_is_nan(row['substrate_2_conc']),
specific_activity=check_is_float(row['specific_activity']),
conversion=check_is_float(row['conversion']),
conversion_time=check_is_float(row['conversion_time']),
categorical=check_is_nan(row['categorical']),
binary=binary,
selectivity=check_is_nan(row['selectivity']),
auto_generated=auto_gen)
activity.save()
print('..done')
def create_paper():
form = PaperInfo()
if 'doi' in session:
doi = session['doi']
else:
doi = ''
if form.validate_on_submit() == False:
user = User.objects(id=current_user.id)[0]
papers = Paper.objects(doi__iexact=doi)
if len(papers) != 0:
paper = papers[0]
if paper.owner == user or user.has_role('super_contributor'):
flash("Paper already in the database", 'success')
return redirect(
url_for("biocatdb.submission_main_page",
paper_id=paper.id))
elif paper.owner == None:
flash(
'Paper already in database, self-assign the paper to add data',
'warning')
return redirect(url_for("biocatdb.papers_that_need_data"))
elif paper.owner != user and not user.has_role(
'super_contributor'):
flash(
"Paper already in the database and is assigned to another user",
'danger')
else:
flash("error", 'danger')
return redirect(url_for("biocatdb.launch_add_paper"))
else:
title, authors_list, journal, date, cite_mini = papers_crossref.get_metadata_from_crossref(
doi)
if cite_mini == '':
title, authors_list, journal, date, cite_mini = papers_functions.query_pubmed(
doi)
form.title.data = title
form.authors.data = papers_functions.list_to_string(authors_list)
form.journal.data = journal
form.date.data = date
form.short_cit.data = cite_mini
form.doi.data = doi.replace(' ', '').lower()
can_self_assign = papers_functions.can_self_assign(user)
if cite_mini == '':
flash(
"Paper not found in crossref, pubmed or db, please add manually",
'fail')
else:
flash("Paper found, please check information", 'success')
return render_template(
'add_paper_workflow/edit_paper_information.html',
form=form,
can_self_assign=can_self_assign)
else:
user = User.objects(id=current_user.id)[0]
session.pop('doi')
if form.self_assign.data == True:
owner = user
else:
owner = None
new_paper = Paper(doi=form.doi.data.replace(' ', '').lower(),
short_citation=form.short_cit.data,
title=form.title.data,
html='https://doi.org/' + form.doi.data,
journal=form.journal.data,
date=form.date.data,
tags=form.tags.data.split(', '),
authors=form.authors.data.split(', '),
owner=owner,
added_by=user,
status='Data required')
new_paper.save()
flash("Paper saved", 'success')
if owner == user:
return redirect(
url_for("biocatdb.submission_main_page",
paper_id=new_paper.id))
else:
return redirect(url_for("biocatdb.launch_add_paper"))
