def family_from_indiv_id_list(indiv_id_list, project_id, family_id):
indivs = [
Individual({
'project_id': project_id,
'family_id': family_id,
'indiv_id': indiv_id
}) for indiv_id in indiv_id_list
]
family = Family({'project_id': project_id, 'family_id': family_id})
for indiv in indivs:
family.add_individual(indiv)
return family
def get_individuals_from_fam_file(fam_file, project_id='.'):
"""
Returns a list of individuals from a FAM file
"""
individuals = []
for line in fam_file:
try:
# ignore these rows
if line == '' or line.startswith('#'): continue
fields = line.strip('\n').split('\t')
indiv_id = slugify(fields[1], separator='_', replace_dot=True)
family_id = slugify(fields[0], separator='_', replace_dot=True)
paternal_id = slugify(fields[2], separator='_', replace_dot=True)
if paternal_id == "0": paternal_id = "."
maternal_id = slugify(fields[3], separator='_', replace_dot=True)
if maternal_id == "0": maternal_id = "."
gender = 'unknown'
if fields[4] == '2' or fields[4].upper().startswith('F'):
gender = 'female'
elif fields[4] == '1' or fields[4].upper().startswith('M'):
gender = 'male'
affected_status = 'unknown'
if fields[5] == '2' or fields[5].upper().startswith('A'):
affected_status = 'affected'
elif fields[5] == '1' or fields[5].upper().startswith('U'):
affected_status = 'unaffected'
except Exception as e:
raise ValueError(
"Couldn't parse line: %(line)s. Fields: %(fields)s. exception: %(e)s"
% locals())
indiv = Individual(
indiv_id,
project_id=project_id,
family_id=family_id,
paternal_id=paternal_id,
maternal_id=maternal_id,
gender=gender,
affected_status=affected_status,
)
individuals.append(indiv)
return individuals
def get_control_cohort(self, population):
indiv_id_list = self.get_individuals_for_family(population, 'control_cohort')
individuals = [Individual(indiv_id, affected_status='affected') for indiv_id in indiv_id_list]
cohort = Cohort('control_cohort', individuals, project_id=population)
return cohort