def rmsd_to_reference(self,
temp_target_ids,
ref_pdb,
pept_chain,
ref_pept_chid=None,
align_mth='SW',
alignment=None,
path=None,
pept_align_kwargs={},
target_align_kwargs={}):
"""
Arguments:
ref_pdb -- str; pdb code of reference structure.
pept_chain -- str; peptide chain name (template).
ref_pept_chain -- str; optional. If set, appropriate chain is picked from reference structure. Otherwise alignment agains all chains is calculated.
align_mth -- str; name of aligning method to be used. See cabsDock.align documentation for more information.
alignment -- str; path to csv alignment file. None by default. If so -- no alignment is loaded. Otherwise target protein is not aligned, instead alignemnt from file is loaded.
path -- str; path to working directory in which alignment is to be saved. None by default. If so -- no file is created.
pept_align_kwargs -- dict of kwargs to be passed to aligning method while aligning peptide.
target_align_kwargs -- as above, but used when aligning target protein.
"""
mth = AbstractAlignMethod.get_subclass_dict()[align_mth]
ref_stc = Pdb(ref_pdb, selection='name CA and not HETERO').atoms
# aligning peptide
if ref_pept_chid is None:
temp_pept = self.template.select(
'name CA and not HETERO and chain %s' % pept_chain)
ref_pept, temp_pept = [
Atoms(arg=list(i)) for i in zip(*mth.execute(
ref_stc, temp_pept, short=True, **pept_align_kwargs))
]
ref_pept_chs = set([i.chid for i in ref_pept.atoms])
if len(ref_pept_chs) > 1:
raise ValueError("Peptide aligned to more tha one chain")
ref_pept_chid = max(ref_pept_chs)
else:
ref_pept = ref_stc.atoms.select('NAME CA and CHAIN %s' %
ref_pept_chid)
#aligning target
if not alignment:
# choosing remaining chains but the one aligned with peptide
ref_target = ref_stc.select('not CHAIN %s' % ref_pept_chid)
temp_target = self.template.select(
"CHAIN %s" % " or CHAIN ".join(temp_target_ids))
# aligning target to reference not-peptide
ref_target_ids = tuple(ref_target.list_chains().keys())
mtch_mtx = numpy.zeros((len(ref_target_ids), len(temp_target_ids)),
dtype=int)
algs = {}
key = 1
for n, rch in enumerate(ref_target_ids):
for m, tch in enumerate(temp_target_ids):
ref = ref_stc.select(
'name CA and not HETERO and chain %s' % rch)
tmp = self.template.select(
'name CA and not HETERO and chain %s' % tch)
if 0 in (len(ref), len(tmp)): continue #??? whai?
try:
algs[key] = mth.execute(ref, tmp)
except AlignError:
continue
mtch_mtx[n, m] = key
key += 1
# joining cabs chains
pickups = []
for n, refch in enumerate(mtch_mtx):
inds = numpy.nonzero(refch)
pickups.extend(refch[inds])
mtch_mtx[n + 1:, inds] = 0
best_alg = reduce(operator.add, [algs.get(k, ()) for k in pickups])
else:
with open(alignment) as f:
best_alg = load_csv(f, ref_stc, self.template)
#saving alignment
if path and not alignment:
save_csv(path, ('ref', 'cabs'), best_alg)
save_fasta(path.replace('csv', 'fasta'), ('ref', 'cabs'),
(self.template, ref_stc), best_alg)
ref_target_mers, temp_target_mers = zip(*best_alg)
structure = Atoms(arg=list(ref_target_mers))
template_aligned = Atoms(arg=list(temp_target_mers))
peptide = numpy.array(ref_pept.to_matrix())
def rmsd(m1, m2, length):
return np.sqrt(np.sum((m1 - m2)**2) / length)
self.align_to(structure, template_aligned=template_aligned)
models_peptide_traj = self.select("chain " + pept_chain)
peptide_length = len(models_peptide_traj.template)
models_peptide = models_peptide_traj.coordinates.reshape(
-1, peptide_length, 3)
result = np.zeros(len(models_peptide))
for i, h in zip(range(len(models_peptide)), self.headers):
result[i] = rmsd(models_peptide[i], peptide, peptide_length)
h.rmsd = result[i]
return result
