def group(self, ifes, inters):
"""Group chains from the same pdb into structured units. This will place
structured chains into their own groups, but allow a non structured
chain to be grouped with it if present. Non structured chains that have
no interacting partners (singleton unstructured chains) are also
present in the output.
:ifes: A list of chain objects from nr.chains.Info.load.
:interactions: A dictionary of interactions of the form produced by
nr.chains.Info.cross_chain_interactions.
:returns: A list of ife groups formed from the given ifes and
interactions.
"""
groups = {}
for ife in ifes:
groups[ife.id] = IfeGroup(ife)
same, rest = self.parition_interactions(ifes, inters)
# If both groups are structured or not structured and the ifes can be
# joined then they should be merged into one group and both ifes should
# reflect this. Basically connections between these types of ifes
# are transitive.
for ife1, ife2 in self.joinable(same, ifes):
current = groups[ife1.id]
current.merge(groups[ife2.id])
linked = set(groups[ife1.id].chains() + groups[ife2.id].chains())
for ife in linked:
groups[ife.id] = current
# Here if we are merging an unstructured into a structured we only
# should update the structured one, the unstructured chain may be part
# of many different structured ifes. Basically connections are not
# transitive across types of groups.
structured = [ife for ife in ifes if ife.is_structured]
unstructured = [ife for ife in ifes if not ife.is_structured]
strip = set()
for ife1, ife2 in self.joinable(rest, structured, unstructured):
groups[ife1.id].merge(groups[ife2.id])
strip.add(groups[ife2.id].id)
# There will be duplicated groups so we must get only the unique ones.
# We sort to ensure that the ordering is consistent.
return sorted(set(g for g in groups.values() if g.id not in strip))
def test_uses_structured_only_in_id(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=5, model=10),
IfeChain(pdb='0111', chain='C', internal=0, model=0),
IfeChain(pdb='0111', chain='D', internal=6, model=10))
self.assertEquals('0111|10|D+0111|10|A', val.id)
def test_has_an_id(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=5, model=10),
IfeChain(pdb='0111', chain='C', internal=0, model=10))
self.assertEquals('0111|10|A', val.id)
def test_knows_how_many_chains_in_it(self):
val = IfeGroup(IfeChain(chain='A', internal=4),
IfeChain(chain='B', internal=0))
self.assertEquals(2, len(val))
def test_knows_if_structured_if_has_no_structured_chain(self):
val = IfeGroup(IfeChain(chain='A', internal=4),
IfeChain(chain='B', internal=0))
self.assertFalse(val.is_structured)
def test_uses_name_to_tiebreak_integral_chain(self):
val = IfeGroup(IfeChain(pdb='0111', chain='C', length=10, internal=0,
model=3),
IfeChain(pdb='0111', chain='A', length=10, internal=0,
model=3))
self.assertEquals('0111|3|A', val.integral.id)
def test_uses_length_as_tiebreak(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', length=10, internal=50,
model=10),
IfeChain(pdb='0111', chain='C', length=50, internal=50,
model=10))
self.assertEquals('0111|10|C', val.integral.id)
def test_uses_most_bp_chain_as_integral(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=4, model=1),
IfeChain(pdb='0111', chain='C', internal=0, model=1))
self.assertEquals('0111|1|A', val.integral.id)
def test_getattr_complains_if_missing_key(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=4,
full_length=5))
self.assertRaises(AttributeError, lambda: val.bob)
def test_dispatches_model_to_integral(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=100,
full_length=5, model=2),
IfeChain(pdb='0111', chain='B', internal=10,
full_length=5, model=10))
self.assertEquals(2, val.model)
def test_dispatches_bps_to_integral(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=100,
full_length=5, bps=120),
IfeChain(pdb='0111', chain='B', internal=10,
full_length=5, bps=12))
self.assertEquals(120, val.bps)
def test_dispatches_pdb_to_integral(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=4,
full_length=5))
self.assertEquals('0111', val.pdb)
def test_duplicate_additions_do_nothing(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=4, model=4),
IfeChain(pdb='0111', chain='C', internal=0, model=4))
val.add(IfeChain(pdb='0111', chain='A', internal=4, model=4))
self.assertEquals('0111|4|A+0111|4|C', val.id)
def test_if_no_structured_in_id_uses_all(self):
val = IfeGroup(IfeChain(pdb='0111', chain='A', internal=2, model=1),
IfeChain(pdb='0111', chain='C', internal=0, model=2),
IfeChain(pdb='0111', chain='D', internal=2, model=3))
self.assertEquals('0111|1|A+0111|3|D+0111|2|C', val.id)