def add_curated(interactions: pd.DataFrame,
interaction_curated: pd.DataFrame) -> pd.DataFrame:
interactions.rename(index=str,
columns={
'uniprot_1': 'partner_a',
'uniprot_2': 'partner_b'
},
inplace=True)
interactions_curated_normalized = normalize_interactions(
interaction_curated, 'partner_a', 'partner_b')
curated_duplicated = interactions_curated_normalized[
interactions_curated_normalized.duplicated(['partner_a', 'partner_b'],
keep=False)]
if not curated_duplicated.empty:
print('WARNING - Some curated interactions are duplicated.')
print(interaction_curated.iloc[curated_duplicated.sort_values(
['partner_a', 'partner_b']).index.values].to_csv(index=False))
interactions.rename(index=str,
columns={
'protein_1': 'partner_a',
'protein_2': 'partner_b'
},
inplace=True)
interaction_curated_result = merge_interactions.merge_iuphar_other_and_curated_interactions(
interactions, interaction_curated)
return interaction_curated_result
def _drop_duplicates(iuphar_filtered_cellphone_format: pd.DataFrame) -> pd.DataFrame:
interactions_normalized = normalize_interactions(iuphar_filtered_cellphone_format, 'uniprot_1',
'uniprot_2')
interactions_duplicated = interactions_normalized[
interactions_normalized.duplicated(['uniprot_1', 'uniprot_2'])]
# if not interactions_duplicated.empty:
# print('WARNING: SOME IUPHAR INTERACTIONS ARE DUPLICATED')
# print(interactions_duplicated.sort_values(['uniprot_1', 'uniprot_2']).to_csv(index=False))
iuphar_procesed = interactions_normalized.drop_duplicates(['uniprot_1', 'uniprot_2'])
return iuphar_procesed
def merge_iuphar_other_and_curated_interactions(
iuphar_other_interactions: pd.DataFrame,
curated_interactions: pd.DataFrame) -> pd.DataFrame:
all_interactions = iuphar_other_interactions.append(curated_interactions)
all_interactions.reset_index(inplace=True, drop=True)
normalized_interactions = normalize_interactions(all_interactions,
'partner_a', 'partner_b')
duplicated_interactions = normalized_interactions[
normalized_interactions.duplicated(['partner_a', 'partner_b'],
keep=False)]
unique_interactions = duplicated_interactions.drop_duplicates(
['partner_a', 'partner_b'])
def merge_values(interaction: pd.Series) -> pd.Series:
interaction = interaction.copy()
duplicated = duplicated_interactions[
(duplicated_interactions['partner_a'] == interaction['partner_a'])
&
(duplicated_interactions['partner_b'] == interaction['partner_b'])]
if not duplicated[duplicated['annotation_strategy'] ==
'curated'].empty:
interaction = duplicated[duplicated['annotation_strategy'] ==
'curated'].iloc[0]
if duplicated['iuphar'].any():
interaction['iuphar'] = True
return interaction
merged_duplicated_interactions = unique_interactions.apply(merge_values,
axis=1)
non_repeated_interactions = normalized_interactions.drop_duplicates(
['partner_a', 'partner_b'], keep=False)
interactions_merged = non_repeated_interactions.append(
merged_duplicated_interactions, sort=True, ignore_index=True)
interactions_merged.fillna({'iuphar': False}, inplace=True)
interactions_merged = interactions_merged.astype({'iuphar': bool})
return interactions_merged
def generate_interactions(
proteins: str,
genes: str,
complex: str,
user_interactions: Optional[str],
user_interactions_only: bool,
result_path: str,
fetch_imex: bool,
fetch_iuphar: bool,
project_name: str,
) -> None:
if user_interactions_only and not user_interactions:
raise Exception('You need to set --user-interactions parameter')
output_path = utils.set_paths(result_path, project_name)
downloads_path = utils.set_paths(
utils.set_paths(result_path, project_name), 'downloads')
proteins = utils.read_data_table_from_file(proteins)
genes = utils.read_data_table_from_file(genes)
complexes = utils.read_data_table_from_file(complex)
if not user_interactions_only:
raw_imex = get_imex.call(genes, downloads_path, fetch_imex)
interactions_to_remove = utils.read_data_table_from_file(
os.path.join(data_dir, 'sources/excluded_interaction.csv'))
interaction_curated = utils.read_data_table_from_file(
os.path.join(data_dir, 'sources/interaction_curated.csv'))
if user_interactions:
separator = _get_separator(os.path.splitext(user_interactions)[-1])
user_interactions = pd.read_csv(user_interactions, sep=separator)
user_interactions['partner_a'] = user_interactions['partner_a'].apply(
lambda x: str(x).strip())
user_interactions['partner_b'] = user_interactions['partner_b'].apply(
lambda x: str(x).strip())
user_interactions['annotation_strategy'] = 'user_curated'
if not 'protein_name_a' in user_interactions.columns:
user_interactions['protein_name_a'] = ''
if not 'protein_name_b' in user_interactions.columns:
user_interactions['protein_name_b'] = ''
result_columns = [
'partner_a', 'partner_b', 'protein_name_a', 'protein_name_b',
'annotation_strategy', 'source'
]
if not user_interactions_only:
print('Parsing IMEX file')
imex_interactions = parse_interactions_imex(raw_imex, proteins, genes)
print('Getting iuphar data')
raw_iuphar = get_iuphar.call(downloads_path, fetch_iuphar)
print('Generating iuphar interactions')
iuphar_interactions = parse_iuphar_guidetopharmacology.call(
raw_iuphar, genes, proteins)
print('Merging iuphar/imex')
merged_interactions = merge_iuphar_imex_interactions(
iuphar_interactions, imex_interactions)
print('Removing complex interactions')
no_complex_interactions = only_noncomplex_interactions(
merged_interactions, complexes)
print('Removing selected interactions')
clean_interactions = remove_interactions_in_file(
no_complex_interactions, interactions_to_remove)
print('Adding curated interaction')
interactions_with_curated = add_curated(clean_interactions,
interaction_curated)
result = tools_helper.normalize_interactions(
interactions_with_curated.append(user_interactions,
ignore_index=True,
sort=False), 'partner_a',
'partner_b').drop_duplicates(['partner_a', 'partner_b'],
keep='last')
else:
result = tools_helper.normalize_interactions(user_interactions, 'partner_a', 'partner_b') \
.drop_duplicates(['partner_a', 'partner_b'], keep='last')
result[result_columns].sort_values(['partner_a', 'partner_b']).to_csv(
'{}/interaction_input.csv'.format(output_path), index=False)