def genes2phens(p2entrez_file, e2z=None):
ensg2z = e2z if e2z else ut.load_dict_sets(ut.proj_path('convert', 'Hs2Hs_entrez.tab'))
def dict_sets_rem_prefix(d, sep):
d = dict([(k,set([vi.split(sep)[1] for vi in v])) for k,v in d.items()])
return d
p2z = dict_sets_rem_prefix(ut.load_dict_sets(p2entrez_file), ":")
return ut.compose_dict_sets(ensg2z, ut.dict_inverse_sets(p2z))
def convert_dict(fromtype, totype):
"""
First looks for single conversion step. If not found, splits it up.
Returns None if not necessary or not found.
"""
conv1 = convert_dict_single(fromtype, totype)
if conv1:
return conv1
else:
# If we made it here, try first converting to second species,
# then looking for other conversion.
conv1 = convert_dict_single(fromtype, totype[:2])
conv2 = convert_dict_single(totype[:2], totype)
if conv1 and conv2:
return ut.compose_dict_sets(conv1,conv2)
def convert_dict(fromtype, totype):
"""
First looks for single conversion step. If not found, splits it up.
Returns None if not necessary or not found.
"""
conv1 = convert_dict_single(fromtype, totype)
if conv1:
return conv1
else:
# If we made it here, try first converting to second species,
# then looking for other conversion.
conv1 = convert_dict_single(fromtype, totype[:2])
conv2 = convert_dict_single(totype[:2], totype)
if conv1 and conv2:
return ut.compose_dict_sets(conv1, conv2)
def orth_indices(sp_base, sp_target, prot_list, remove_multi_base):
"""
Using appropriate orthology, take a list of target species gene ids
(corresponding to rows in the target species score matrix), and
return a dict mapping base species gene ids to (sets of) indices in that
list and therefore to (sets of) row/column indices in the square
interaction score matrix.
"""
targ2inds = dict([(k,set([v]))
for k,v in ut.list_inv_to_dict(prot_list).items()])
if sp_base == sp_target:
return targ2inds
else:
base2targ = orth.odict(sp_base, sp_target)
if remove_multi_base:
base2targ = remove_multi_keys(base2targ)
base2inds = ut.compose_dict_sets(base2targ, targ2inds)
base2inds = dict([(k,v) for k,v in base2inds.items() if len(v)>0])
return base2inds
def orth_indices(sp_base, sp_target, prot_list, remove_multi_base):
"""
Using appropriate orthology, take a list of target species gene ids
(corresponding to rows in the target species score matrix), and
return a dict mapping base species gene ids to (sets of) indices in that
list and therefore to (sets of) row/column indices in the square
interaction score matrix.
"""
targ2inds = dict([(k,set([v]))
for k,v in ut.list_inv_to_dict(prot_list).items()])
if sp_base == sp_target:
return targ2inds
else:
base2targ = orth.odict(sp_base, sp_target)
if remove_multi_base:
base2targ = remove_multi_keys(base2targ)
base2inds = ut.compose_dict_sets(base2targ, targ2inds)
base2inds = dict([(k,v) for k,v in base2inds.items() if len(v)>0])
return base2inds