def call(gene_expanded, output_columns):
filters.remove_not_defined_columns(gene_expanded, output_columns)
gene_expanded.drop(['id_gene', 'protein_id'], axis=1, inplace=True)
gene_expanded.rename(index=str, columns={'name': 'uniprot'}, inplace=True)
return gene_expanded
def add(self, interactions):
interaction_df = self.blend_dataframes(interactions, ['partner_a', 'partner_b'],
self.database_manager.get_repository('multidata').get_all_name_id(),
'name', 'multidata')
filters.remove_not_defined_columns(interaction_df,
self.database_manager.get_column_table_names('interaction_table'))
interaction_df.to_sql(name='interaction_table', if_exists='append', con=self.database_manager.database.engine,
index=False, chunksize=50)
def call(interactions_expanded: pd.DataFrame) -> pd.DataFrame:
interactions_expanded.rename(index=str, columns={'name_1': 'multidata_name_1', 'name_2': 'multidata_name_2'},
inplace=True)
filters.remove_not_defined_columns(interactions_expanded,
['comments_interaction', 'entry_name_1', 'entry_name_2', 'multidata_name_1',
'multidata_name_2', 'score_1', 'score_2', 'source'])
interactions_extended = dataframe_format.bring_columns_to_start(
['multidata_name_1', 'entry_name_1', 'multidata_name_2',
'entry_name_2'], interactions_expanded)
return interactions_extended
def _only_uniprots_in_df(uniprots_df, inweb_interactions):
inweb_cellphone = pd.merge(inweb_interactions,
uniprots_df,
left_on=['protein_1'],
right_on=['uniprot'],
how='inner')
remove_not_defined_columns(inweb_cellphone,
inweb_interactions.columns.values)
inweb_cellphone = pd.merge(inweb_cellphone,
uniprots_df,
left_on=['protein_2'],
right_on=['uniprot'],
how='inner')
remove_not_defined_columns(inweb_cellphone,
inweb_interactions.columns.values)
# Prevents duplicated interactions if any uniprot is duplicated in uniprots_df or intaractions
inweb_cellphone = inweb_cellphone[inweb_cellphone.duplicated() == False]
return remove_not_defined_columns(inweb_cellphone,
inweb_interactions.columns.values)
def add(self, genes: pd.DataFrame):
query_multidatas = self.database_manager.database.session.query(
Protein.id_protein, Multidata.name).join(Multidata)
multidatas = pd.read_sql(query_multidatas.statement,
self.database_manager.database.session.bind)
genes = self._blend_multidata(genes, ['name'], multidatas)
genes.rename(index=str,
columns={'id_protein': 'protein_id'},
inplace=True)
genes = filters.remove_not_defined_columns(
genes, self.database_manager.get_column_table_names('gene'))
genes.to_sql(name='gene',
if_exists='append',
con=self.database_manager.database.engine,
index=False,
chunksize=50)
def add(self, complexes):
"""
Uploads complex data from csv.
- Creates new complexes in Multidata table
- Creates reference in Complex table
- Creates complex composition to define complexes.
"""
if complexes.empty:
return
existing_complexes = self.database_manager.database.session.query(Multidata.name).all()
existing_complexes = [c[0] for c in existing_complexes]
proteins = self.database_manager.database.session.query(Multidata.name, Multidata.id_multidata).join(
Protein).all()
proteins = {p[0]: p[1] for p in proteins}
# Get complex composition info
complete_indices = []
incomplete_indices = []
missing_proteins = []
complex_map = {}
for index, row in complexes.iterrows():
missing = False
protein_id_list = []
for protein in ['protein_1', 'protein_2',
'protein_3', 'protein_4']:
if not pd.isnull(row[protein]):
protein_id = proteins.get(row[protein])
if protein_id is None:
missing = True
missing_proteins.append(row[protein])
else:
protein_id_list.append(protein_id)
if not missing:
complex_map[row['name']] = protein_id_list
complete_indices.append(int(index))
else:
incomplete_indices.append(index)
if len(incomplete_indices) > 0:
core_logger.warning('MISSING PROTEINS:')
for protein in missing_proteins:
core_logger.warning('MISSING PROTEINS:')(protein)
core_logger.warning('COMEPLEXES WITH MISSING PROTEINS:')
core_logger.warning(complexes.iloc[incomplete_indices, :]['name'])
# Insert complexes
if not complexes.empty:
# Remove unwanted columns
removal_columns = list(
[x for x in complexes.columns if 'protein_' in x or 'Name_' in x or 'Unnamed' in x])
# removal_columns += ['comments']
complexes.drop(removal_columns, axis=1, inplace=True)
# Remove rows with missing complexes
complexes = complexes.iloc[complete_indices, :]
# Convert ints to bool
bools = ['receptor', 'other', 'secreted_highlight', 'transmembrane', 'secreted',
'peripheral']
complexes[bools] = complexes[bools].astype(bool)
# Drop existing complexes
complexes = complexes[complexes['name'].apply(
lambda x: x not in existing_complexes)]
multidata_df = filters.remove_not_defined_columns(complexes.copy(),
self.database_manager.get_column_table_names(
'multidata_table'))
multidata_df = self._add_complex_optimitzations(multidata_df)
multidata_df.to_sql(name='multidata_table', if_exists='append', con=self.database_manager.database.engine,
index=False, chunksize=50)
# Now find id's of new complex rows
new_complexes = self.database_manager.database.session.query(Multidata.name, Multidata.id_multidata).all()
new_complexes = {c[0]: c[1] for c in new_complexes}
# Build set of complexes
complex_set = []
complex_table = []
for complex_name in complex_map:
complex_id = new_complexes[complex_name]
for protein_id in complex_map[complex_name]:
complex_set.append((complex_id, protein_id, len(complex_map[complex_name])))
complex_table.append({'complex_multidata_id': complex_id, 'name': complex_name})
# Insert complex composition
complex_set_df = pd.DataFrame(complex_set,
columns=['complex_multidata_id', 'protein_multidata_id', 'total_protein'])
complex_table_df = pd.DataFrame(complex_table)
complex_table_df = pd.merge(complex_table_df, complexes, on='name')
filters.remove_not_defined_columns(complex_table_df,
self.database_manager.get_column_table_names('complex_table'))
complex_table_df.to_sql(
name='complex_table', if_exists='append',
con=self.database_manager.database.engine, index=False, chunksize=50)
complex_set_df.to_sql(
name='complex_composition_table', if_exists='append',
con=self.database_manager.database.engine, index=False, chunksize=50)
def process_protein_data(proteins: pd.DataFrame,
protein_columns: list) -> pd.DataFrame:
proteins = proteins.copy()
proteins = filters.remove_not_defined_columns(proteins,
protein_columns + ['name'])
return proteins
def process_multidata_data(proteins: pd.DataFrame,
multidata_columns: list) -> pd.DataFrame:
multidata_proteins = proteins.copy()
return filters.remove_not_defined_columns(multidata_proteins,
multidata_columns)
def call(genes: pd.DataFrame, gene_columns: list) -> pd.DataFrame:
genes.rename(index=str, columns={'uniprot': 'name'}, inplace=True)
filters.remove_not_defined_columns(genes, gene_columns + ['name'])
return genes
