def check_if_there_is_allele_with_4_diff_values(fam_d, alleles_names, invalid_cases):
"""
if there are no parents, and there isn't an allele with 4 different values in the children,
the algorithm could not be executed
(because when no parents, we rely on the assumption that there are 4 different in one allele, at least)
it's not an "invalid" input, but we reject the family because we can not analyze it
"""
four_different_values = False # flag to know if there is allele with 4 different values
for al_name in alleles_names:
if four_different_values: # if there is allele with 4 diff, do not need to check more
return
alleles_values = Als()
for child in fam_d: # no parents, according the condition in the call to this function
child_alleles = fam_d[child][al_name]
if any(child_alleles): # merge only if there is data in 'child_alleles'
# merge the values of the current alleles_names _child_ to the values of the other children in this allele
# e.g. : alleles_values: [02, 03]. child_alleles: [02:01, 04]. so the merging: [02, 03, 04]
# note: the merging may save the low-res values (02 instead of 02:01),
# but it's not matter, because just need the values amount
alleles_values = alleles_values.merge(child_alleles)
# after go over on all the children values in the current allele, check if there are 4 values
if len(alleles_values) == 4:
four_different_values = True
if not four_different_values:
invalid_cases.append(('6', 'All')) # no parents, and no alleles_names with 4 diff values (algorithm can not executed)
def create_merged_allele_values(children, al_name):
"""
create a list (Als) with all the different value in children data about specific allele
for example: if _child_ 1 has A : [02:01, 03:04], _child_ 2: A: [05, 03], _child_ 3: A: [02, 07],
so lst will be [02:01, 03:04, 05, 07]
:param children: children dict
:param al_name: name of specific allele
"""
lst = Als()
for child in children:
lst = lst.merge(children[child][al_name])
return lst
def check_too_much_alleles(fam_d, alleles_names, invalid_cases):
"""
check if there are too much alleles_names in the family (more than 4 in an allele)
"""
for al_name in alleles_names:
lst = Als()
for fam_member in fam_d: # [F, M, 1 ...]
if any(fam_d[fam_member][al_name]): # not empty
# lst = fam_d[fam_member][al_name].merge(lst)
lst = lst.merge(fam_d[fam_member][al_name])
if len(lst) > 4:
invalid_cases.append(('4', 'All')) # Too many alleles_names