def test_hbond_with_selections():
"""
When selection1 and selection2 is defined, no hydrogen bonds outside
of this boundary should be found. Also, hbond should respect the
selection type.
"""
stack = load_structure(join(data_dir, "1l2y.mmtf"))
selection1 = (stack.res_id == 3) & (stack.atom_name == 'O') # 3TYR BB Ox
selection2 = stack.res_id == 7
# backbone hbond should be found if selection1/2 type is both
triplets, mask = struc.hbond(stack, selection1, selection2,
selection1_type="both")
assert len(triplets) == 1
assert triplets[0][0] == 116
assert triplets[0][2] == 38
# backbone hbond should be found if selection1 is acceptor and
# selection2 is donor
triplets, mask = struc.hbond(stack, selection1, selection2,
selection1_type="acceptor")
assert len(triplets) == 1
assert triplets[0][0] == 116
assert triplets[0][2] == 38
# no hbond should be found,
# because the backbone oxygen cannot be a donor
triplets, mask = struc.hbond(stack, selection1, selection2,
selection1_type="donor")
assert len(triplets) == 0
def test_hbond_structure(pdb_id):
file_name = join(data_dir("structure"), pdb_id + ".mmtf")
array = load_structure(file_name)
# Only consider amino acids for consistency
# with bonded hydrogen detection in MDTraj
array = array[..., struc.filter_amino_acids(array)]
if isinstance(array, struc.AtomArrayStack):
# For consistency with MDTraj 'S' cannot be acceptor element
# https://github.com/mdtraj/mdtraj/blob/master/mdtraj/geometry/hbond.py#L365
triplets, mask = struc.hbond(array, acceptor_elements=("O", "N"))
else:
triplets = struc.hbond(array, acceptor_elements=("O", "N"))
# Save to new pdb file for consistent treatment of inscode/altloc
# im MDTraj
temp = NamedTemporaryFile("w+", suffix=".pdb")
save_structure(temp.name, array)
# Compare with MDTraj
import mdtraj
traj = mdtraj.load(temp.name)
temp.close()
triplets_ref = mdtraj.baker_hubbard(traj, freq=0, periodic=False)
# Both packages may use different order
# -> use set for comparison
triplets_set = set([tuple(tri) for tri in triplets])
triplets_ref_set = set([tuple(tri) for tri in triplets_ref])
assert triplets_set == triplets_ref_set
def test_hbond_single_selection():
"""
If only selection1 or selection2 is defined, hbond should run
against all other atoms as the other selection.
"""
stack = load_structure(join(data_dir("structure"), "1l2y.mmtf"))
selection = (stack.res_id == 2) & (stack.atom_name == "O") # 2LEU BB Ox
triplets, mask = struc.hbond(stack, selection1=selection)
assert len(triplets) == 2
triplets, mask = struc.hbond(stack, selection2=selection)
assert len(triplets) == 2
def test_hbond_total_count():
"""
With the standart Baker & Hubbard criterion,
1l2y should have 28 hydrogen bonds with a frequency > 0.1
(comparision with MDTraj results)
"""
stack = load_structure(join(data_dir("structure"), "1l2y.mmtf"))
triplets, mask = struc.hbond(stack)
freq = struc.hbond_frequency(mask)
assert len(freq[freq >= 0.1]) == 28
def test_hbond_same_res():
"""
Check if hydrogen bonds in the same residue are detected.
At least one of such bonds is present in 1L2Y (1ASN with N-terminus)
(model 2).
"""
stack = load_structure(join(data_dir("structure"), "1l2y.mmtf"))
selection = stack.res_id == 1
# Focus on second model
array = stack[1]
triplets = struc.hbond(array, selection, selection)
assert len(triplets) == 1
def test_hbond_periodicity(translation_vector):
"""
Test whether hydrogen bond identification uses periodic boundary
conditions correctly.
For this purpose a structure containing water is loaded and the
hydrogen bonds are identified.
Then the position of the periodic boundary is changed and it is
expected that all hydrogen bonds are still the same
"""
stack = load_structure(join(data_dir("structure"), "waterbox.gro"))
array = stack[0]
ref_hbonds = struc.hbond(array, periodic=True)
# Put H-bond triplets into as stack for faster comparison with
# set for moved atoms
ref_hbonds = set([tuple(triplet) for triplet in ref_hbonds])
# Move system and put back into box
# -> Equal to move of periodic boundary
array = struc.translate(array, translation_vector)
array.coord = struc.move_inside_box(array.coord, array.box)
hbonds = struc.hbond(array, periodic=True)
hbonds = set([tuple(triplet) for triplet in hbonds])
assert ref_hbonds == hbonds
import matplotlib.pyplot as plt
import biotite.structure as struc
import biotite.structure.io as strucio
import biotite.database.rcsb as rcsb
file_name = rcsb.fetch("2KB1", "mmtf", biotite.temp_dir())
stack = strucio.load_structure(file_name)
# Four identical chains, consider only chain A
chain_a = stack[:, stack.chain_id == "A"]
# Selection for p-helix
p_helix = (chain_a.res_id >= 40) & (chain_a.res_id <= 52)
# Selection for selectivity filter
sf = (chain_a.res_id >= 53) & (chain_a.res_id <= 58)
# Calculate the hydrogen bonds and the frequency of each bond
triplets, mask = struc.hbond(chain_a, selection1=p_helix, selection2=sf)
freq = struc.hbond_frequency(mask)
# Create names of bonds
label = "{d_resid}{d_resnm}-{d_a} -- {a_resid}{a_resnm}-{a_a}"
names = [label.format(
d_resid=chain_a.res_id[donor],
d_resnm=chain_a.res_name[donor],
d_a=chain_a.atom_name[donor],
a_resid=chain_a.res_id[acceptor],
a_resnm=chain_a.res_name[acceptor],
a_a=chain_a.atom_name[acceptor]
) for donor, _, acceptor in triplets]
plt.subplots(figsize=(11,4.5))
plt.bar(names, freq, color=biotite.colors["orange"])
atoms = pairs[0][0] + pairs[0][1] + pairs[1][0] + pairs[1][1]
# Color by element
colors = np.zeros((atoms.array_length(), 3))
colors[atoms.element == "H"] = (0.8, 0.8, 0.8) # gray
colors[atoms.element == "C"] = (0.2, 0.2, 0.2) # darkgray
colors[atoms.element == "N"] = (0.0, 0.0, 0.8) # blue
colors[atoms.element == "O"] = (0.8, 0.0, 0.0) # red
colors[atoms.element == "P"] = (0.0, 0.6, 0.0) # green
graphics.plot_atoms(
ax, atoms, colors, line_width=3, background_color="white"
)
# Plot hydrogen bonds
for purine, pyrimidine in pairs:
pair = purine + pyrimidine
bonds = struc.hbond(pair)
for donor, hydrogen, acceptor in bonds:
hydrogen_coord = pair.coord[hydrogen]
acceptor_coord = pair.coord[acceptor]
x, y, z = zip(hydrogen_coord, acceptor_coord)
ax.plot(x, y, z, linestyle=":", color="gold", linewidth=2)
# Label bases
for pair in pairs:
for base in pair:
label = base.res_name[0][1]
ring_center = struc.centroid(base[
np.isin(base.atom_name, ["N1", "C2", "N3", "C4", "C5", "C6"])
])
x, y, z = ring_center
ax.text(
structure = structure[~struc.filter_solvent(structure)
& ~struc.filter_monoatomic_ions(structure)]
strep_mask = struc.filter_amino_acids(structure)
biotin_mask = (structure.res_name == "BTN")
pymol_obj = ammolite.PyMOLObject.from_structure(structure)
pymol_obj.cartoon("loop", strep_mask)
pymol_obj.set("cartoon_transparency", 0.5)
pymol_obj.color("salmon", strep_mask & (structure.element == "C"))
pymol_obj.color("forest", biotin_mask & (structure.element == "C"))
pymol_obj.zoom(biotin_mask, buffer=5.0)
ammolite.show(PNG_SIZE)
########################################################################
bonds = struc.hbond(structure, strep_mask, biotin_mask)
res_ids = []
for i, (donor, hydrogen, acceptor) in enumerate(bonds):
# Highlight streptavidin residues bound to biotin
if structure.res_name[donor] != "BTN":
res_id = structure.res_id[donor]
res_name = structure.res_name[donor]
elif structure.res_name[acceptor] != "BTN":
res_id = structure.res_id[acceptor]
res_name = structure.res_name[acceptor]
else:
ValueError("The bond is not between streptavidin and biotin")
res_ids.append(res_id)
pymol_obj.show("sticks", (structure.res_id == res_id)
& ~np.isin(structure.atom_name, BACKBONE_ATOMS))
