def test_write_2dsplinelib(self):
"""Test the write_2dsplinelib function"""
mlib = self.get_mdt_library()
restype = mdt.features.ResidueType(mlib)
phi = mdt.features.PhiDihedral(mlib,
bins=mdt.uniform_bins(18, -180.0, 40.0))
psi = mdt.features.PsiDihedral(mlib,
bins=mdt.uniform_bins(18, -180.0, 40.0))
m = self.get_test_mdt(mlib, [restype, phi, psi])
mdt.write_2dsplinelib(open('test.out', 'w'), m)
# Make sure that valid Python code was produced
code = compile(open('test.out').read(), 'test.out', 'exec')
os.unlink('test.out')
def test_write_2dsplinelib(self):
"""Test the write_2dsplinelib function"""
mlib = self.get_mdt_library()
restype = mdt.features.ResidueType(mlib)
phi = mdt.features.PhiDihedral(mlib,
bins=mdt.uniform_bins(18, -180.0, 40.0))
psi = mdt.features.PsiDihedral(mlib,
bins=mdt.uniform_bins(18, -180.0, 40.0))
m = self.get_test_mdt(mlib, [restype, phi, psi])
with open('test.out', 'w') as fh:
mdt.write_2dsplinelib(fh, m)
# Make sure that valid Python code was produced
with open('test.out') as fh:
_ = compile(fh.read(), 'test.out', 'exec')
os.unlink('test.out')
m = m.reshape(features=(xray, restyp, psi, phi),
offset=(0, 0, 0, 0),
shape=(1, -2, -1, -1))
# Let's get rid of the resolution variable from the output MDT table:
m = m.integrate(features=(restyp, psi, phi))
# Process the raw histograms to get appropriate pdf 1D splines for restraints:
# Start by smoothing with a uniform prior (equal weight when 10 points per bin),
# producing a normalized distribution that sums to 1 (not a pdf when dx != 1):
m = m.smooth(dimensions=2, weight=10)
# Normalize it to get the true pdf (Integral p(x) dx = 1):
# (the scaling actually does not matter, because I am eventually taking the
# log and subtracting the smallest element of the final pdf, so this command
# could be omitted without impact):
m = m.normalize(to_pdf=True, dimensions=2, dx_dy=(5., 5.), to_zero=True)
# Take the logarithm of the smoothed frequencies
# (this is safe: none of bins is 0 because of mdt.smooth()):
m = m.log_transform(offset=0., multiplier=1.)
# Reverse the sign:
m = m.linear_transform(offset=0., multiplier=-1.)
# Offset the final distribution so that the lowest value is at 0:
m = m.offset_min(dimensions=2)
mdt.write_2dsplinelib(file("phipsi.py", "w"), m, density_cutoff=0.1)
