def parse_pharmgkb_line(self, this_line: str, header_items) -> dict:
"""Parse a single line from relationships.tsv and return a dict with data
:param this_line: line from relationship.tsv to parse
:param header_items: header from relationships.tsv
:return: dict with key value containing data
"""
items = this_line.strip().split('\t')
return data_to_dict(header_items, items)
def parse_drug_central_line(this_line: str, header_items: List) -> Dict:
"""Methods processes a line of text from Drug Central.
Args:
this_line: A string containing a line of text.
header_items: A list of header items.
Returns:
item_dict: A dictionary of header items and a processed Drug Central string.
"""
data = this_line.strip().split("\t")
data = [i.replace('"', '') for i in data]
item_dict = data_to_dict(header_items, data)
return item_dict
def make_id_mapping_file(self,
map_file: str,
sep: str = '\t',
pharmgkb_id_col: str = 'PharmGKB Accession Id',
id_key: str = 'Cross-references',
id_sep: str = ',',
id_key_val_sep: str = ':') -> dict:
"""Fxn to parse gene ID mappings or drug ID mapping for PharmGKB ids
This is to parse both genes.tsv and drugs.tsv files
:param map_file: genes.tsv file, containing mappings
:param pharmgkb_id_col: column containing pharmgkb, to be used as key for map
:param sep: separator between columns [\t]
:param id_key: column name that contains ids [Cross-references]
:param id_sep: separator between each id key:val pair [,]
:param id_key_val_sep: separator between key:val pair [:]
:return:
"""
map: dict = defaultdict()
with open(map_file) as f:
header_items = f.readline().split(sep)
if pharmgkb_id_col not in header_items:
raise CantFindPharmGKBKey(
"Can't find PharmGKB id in map file!")
for line in f:
items = line.strip().split(sep)
dat = data_to_dict(header_items, items)
if id_key in dat:
for item in dat[id_key].split(id_sep):
if not item:
continue # not xrefs, skip
item = item.strip(
'\"') # remove quotes around each item
key, value = item.split(id_key_val_sep,
1) # split on first :
if self.key_parsed_ids not in dat:
dat[self.key_parsed_ids] = dict()
dat[self.key_parsed_ids][key] = value
map[dat[pharmgkb_id_col]] = dat
return map