def validate_sequences(sequences, regexp, name, minimum_CG, maximum_CG, maximum_offtarget, immuno):
"""
Here we remove all bad sequences (siRNA) by validators
"""
# filter sequences here by no expensive features
preprocessed = filter(
lambda sequence: all(
[
validate_gc_content(sequence, minimum_CG, maximum_CG),
validate_immuno(sequence, immuno),
validate_thermostability(sequence),
]
),
sequences,
)
# uncomment if debuging
# return {
# name: [{
# "sequence": seq,
# "regexp": int(regexp),
# "offtarget": 0}
# for seq in preprocessed]
# }
# counting offtarget is expensive
with allow_join_result():
offtarget = (
group(offtarget_seed.s(sequence).set(queue="blast") for sequence in preprocessed).apply_async().get()
)
return {
name: [
{"sequence": sequence, "regexp": int(regexp), "offtarget": actual_offtarget}
for sequence, actual_offtarget in izip(preprocessed, offtarget)
if actual_offtarget <= maximum_offtarget
]
}
def test_validate_gc_content_not_in_range(self):
not_in_range = validators.validate_gc_content('ACGT', 52, 70)
self.assertFalse(not_in_range)
def test_validate_gc_content(self):
is_in_range = validators.validate_gc_content('ACGT', 40, 60)
self.assertTrue(is_in_range)
