def _f_process(self, x):
__doc___ = """
returns the squared sum of residuals
objective function is the free energy
"""
self._f[:] = 0.
if self.anal_grad:
self._g[:] = 0.
### &PyArray_Type, &target,
### &PyArray_Type, &source,
### &PyArray_Type, &assign,
### &PyArray_Type, &factor,
### &PyArray_Type, &assign_factor
### Calculate DDG_comp - DDG_exp
if self.pairs:
_f = merge_casedata_ext(self._calc_data, self.ind_case, self.w,
self.ind_rec)
self._f[:] += pair_difference_ext(_f, self.pairidx) + x[-1]
for i in range(self.parms - 1):
_g = merge_casedata_ext(self._gradients[:,
i], self.ind_case,
self.w, self.ind_rec)
self._g[:, i] += pair_difference_ext(_g, self.pairidx)
self._g[:, -1] = 1
else:
self._f[:] += merge_casedata_ext(self._calc_data,
self.ind_case, self.w,
self.ind_rec)
for i in range(self.parms):
self._g[:,
i] += merge_casedata_ext(self._gradients[:, i],
self.ind_case, self.w,
self.ind_rec)
else:
### &PyArray_Type, &target,
### &PyArray_Type, &source,
### &PyArray_Type, &assign,
### &PyArray_Type, &factor,
### &PyArray_Type, &assign_factor
if self.pairs:
_f = merge_casedata_ext(self._calc_data, self.ind_case, self.w,
self.ind_rec)
self._f[:] += pair_difference_ext(_f, self.pairidx) + x[-1]
else:
self._f[:] += merge_casedata_ext(self._calc_data,
self.ind_case, self.w,
self.ind_rec)
def set_functional(self):
if self.parms == 6:
self._gist_functional_ext = gist_functional_6p_ext
elif self.parms == 5:
self._gist_functional_ext = gist_functional_5p_ext
elif self.parms == 4:
self._gist_functional_ext = gist_functional_4p_ext
else:
parms_error(self.parms, self._parms)
### Note, all arrays which are passed to the functionals (such as
### gist_functional_6p_ext), must be DOUBLE (i.e. 32bit floating
### point type in C). This will not checked within the C routine
### (but should be implemented at some point ...).
if self.pairs:
self._exp_data = pair_difference_ext(self.dg.astype(DOUBLE),
self.pairidx)
self._f = np.zeros(self.N_pairs, dtype=DOUBLE)
self._g = np.zeros((self.N_pairs, self._parms), dtype=DOUBLE)
else:
self._exp_data = np.copy(self.dg.astype(DOUBLE))
self._f = np.zeros(self.N_case, dtype=DOUBLE)
self._g = np.zeros((self.N_case, self._parms), dtype=DOUBLE)
self._gradients = np.zeros((self.N_pos, self.parms), dtype=DOUBLE)
self._calc_data = np.zeros(self.N_pos, dtype=DOUBLE)
self._dx = 0.00000001
def _f_process(self, x):
__doc___ = """
returns the squared sum of residuals
objective function is the free energy
"""
self.gist_functional(x)
### x[0] = E_aff
### x[1] = e_co
### x[2] = S_aff
### x[3] = s_co
### x[4] = g_co (Rec)
### x[5] = g_co (Cplx)
### x[6] = g_co (Lig)
### x[7] = C_E
### x[8] = C_S
self._f[:] = 0.
if self.anal_grad:
self._g[:] = 0.
### Complex and Ligand contributions
### &PyArray_Type, &source,
### &PyArray_Type, &assign,
### &PyArray_Type, &factor,
### &PyArray_Type, &assign_factor
if self.pairs:
_f = merge_casedata_ext(self._calc_data_cplx[:, 0],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 0] += pair_difference_ext(_f, self.pairidx)
_f = merge_casedata_ext(self._calc_data_cplx[:, 1],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 1] += pair_difference_ext(_f, self.pairidx)
for i in range(self.parms - 2):
_g = merge_casedata_ext(self._gradients_cplx[:, i, 0],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_g -= merge_casedata_ext(self._gradients_lig[:, i, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, i, 0] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_cplx[:, i, 1],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_g -= merge_casedata_ext(self._gradients_lig[:, i, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, i, 1] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_cplx[:, -3, 0],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
self._g[:, -4, 0] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_lig[:, -3, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, -3, 0] -= pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_cplx[:, -3, 1],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
self._g[:, -4, 1] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_lig[:, -3, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, -3, 1] -= pair_difference_ext(_g, self.pairidx)
else:
self._f[:, 0] = merge_casedata_ext(self._calc_data_cplx[:, 0],
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:, 0] -= merge_casedata_ext(self._calc_data[:, 0],
self.ind_case, self.w,
self.ind_rec)
self._f[:,
0] -= merge_casedata_ext(self._calc_data_lig[:, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 1] = merge_casedata_ext(self._calc_data_cplx[:, 1],
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:, 1] -= merge_casedata_ext(self._calc_data[:, 1],
self.ind_case, self.w,
self.ind_rec)
self._f[:,
1] -= merge_casedata_ext(self._calc_data_lig[:, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
for i in range(self.parms - 2):
self._g[:, i, 0] = merge_casedata_ext(
self._gradients_cplx[:, i, 0], self.ind_case_cplx,
self.w_cplx, self.ind_case_cplx)
self._g[:, i,
0] -= merge_casedata_ext(self._gradients[:, i, 0],
self.ind_case, self.w,
self.ind_rec)
self._g[:, i, 0] -= merge_casedata_ext(
self._gradients_lig[:, i, 0], self.ind_case_lig,
self.w_lig, self.ind_case_lig)
self._g[:, i, 1] = merge_casedata_ext(
self._gradients_cplx[:, i, 1], self.ind_case_cplx,
self.w_cplx, self.ind_case_cplx)
self._g[:, i,
1] -= merge_casedata_ext(self._gradients[:, i, 1],
self.ind_case, self.w,
self.ind_rec)
self._g[:, i, 1] -= merge_casedata_ext(
self._gradients_lig[:, i, 1], self.ind_case_lig,
self.w_lig, self.ind_case_lig)
self._g[:, -4, 0] += merge_casedata_ext(
self._gradients_cplx[:, -3, 0], self.ind_case_cplx,
self.w_cplx, self.ind_case_cplx)
self._g[:, -5,
0] -= merge_casedata_ext(self._gradients[:, -3, 0],
self.ind_case, self.w,
self.ind_rec)
self._g[:, -3,
0] -= merge_casedata_ext(self._gradients_lig[:, -3, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, -4, 1] += merge_casedata_ext(
self._gradients_cplx[:, -3, 1], self.ind_case_cplx,
self.w_cplx, self.ind_case_cplx)
self._g[:, -5,
1] -= merge_casedata_ext(self._gradients[:, -3, 1],
self.ind_case, self.w,
self.ind_rec)
self._g[:, -3,
1] -= merge_casedata_ext(self._gradients_lig[:, -3, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 0] += x[-2]
self._f[:, 1] += x[-1]
self._g[:, -2, 0] = 1
self._g[:, -2, 1] = 0
self._g[:, -1, 0] = 0
self._g[:, -1, 1] = 1
else:
if self.pairs:
_f = merge_casedata_ext(self._calc_data_cplx[:, 0],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 0] += pair_difference_ext(_f, self.pairidx)
_f = merge_casedata_ext(self._calc_data_cplx[:, 1],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 1] += pair_difference_ext(_f, self.pairidx)
else:
self._f[:, 0] = merge_casedata_ext(self._calc_data_cplx[:, 0],
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:, 0] -= merge_casedata_ext(self._calc_data[:, 0],
self.ind_case, self.w,
self.ind_rec)
self._f[:,
0] -= merge_casedata_ext(self._calc_data_lig[:, 0],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 1] = merge_casedata_ext(self._calc_data_cplx[:, 1],
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:, 1] -= merge_casedata_ext(self._calc_data[:, 1],
self.ind_case, self.w,
self.ind_rec)
self._f[:,
1] -= merge_casedata_ext(self._calc_data_lig[:, 1],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:, 0] += x[-2]
self._f[:, 1] += x[-1]
def _f_process(self, x):
__doc__ = """
returns the squared sum of residuals using
multiobjective function optimization with linear
scaling.
objective funtions are energy and entropy
"""
self.gist_functional(x)
self._f[:] = 0.
if self.anal_grad:
self._g[:] = 0.
if self.pairs:
_f = merge_casedata_ext(self._calc_data[:, 0], self.ind_case,
self.w, self.ind_rec)
self._f[:, 0] += pair_difference_ext(_f, self.pairidx) + x[-2]
_f = merge_casedata_ext(self._calc_data[:, 1], self.ind_case,
self.w, self.ind_rec)
self._f[:, 1] += pair_difference_ext(_f, self.pairidx) + x[-1]
for i in range(self.parms - 1):
_g = merge_casedata_ext(self._gradients[:, i,
0], self.ind_case,
self.w, self.ind_rec)
self._g[:, i, 0] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients[:, i,
1], self.ind_case,
self.w, self.ind_rec)
self._g[:, i, 1] += pair_difference_ext(_g, self.pairidx)
self._g[:, -2, 0] = 1
self._g[:, -2, 1] = 0
self._g[:, -1, 0] = 0
self._g[:, -1, 1] = 1
else:
self._f[:, 0] += merge_casedata_ext(self._calc_data[:, 0],
self.ind_case, self.w,
self.ind_rec)
self._f[:, 1] += merge_casedata_ext(self._calc_data[:, 1],
self.ind_case, self.w,
self.ind_rec)
for i in range(self.parms + 1):
self._g[:, i,
0] += merge_casedata_ext(self._gradients[:, i, 0],
self.ind_case, self.w,
self.ind_rec)
self._g[:, i,
1] += merge_casedata_ext(self._gradients[:, i, 1],
self.ind_case, self.w,
self.ind_rec)
else:
if self.pairs:
_f = merge_casedata_ext(self._calc_data[:, 0], self.ind_case,
self.w, self.ind_rec)
self._f[:, 0] += pair_difference_ext(_f, self.pairidx) + x[-2]
_f = merge_casedata_ext(self._calc_data[:, 1], self.ind_case,
self.w, self.ind_rec)
self._f[:, 1] += pair_difference_ext(_f, self.pairidx) + x[-1]
else:
self._f[:, 0] += merge_casedata_ext(self._calc_data[:, 0],
self.ind_case, self.w,
self.ind_rec)
self._f[:, 1] += merge_casedata_ext(self._calc_data[:, 1],
self.ind_case, self.w,
self.ind_rec)
def _f_process(self, x):
__doc___ = """
returns the squared sum of residuals
objective function is the free energy
"""
self.gist_functional(x)
self._f[:] = 0.
if self.anal_grad:
self._g[:] = 0.
### Complex and Ligand contributions
### &PyArray_Type, &source,
### &PyArray_Type, &assign,
### &PyArray_Type, &factor,
### &PyArray_Type, &assign_factor
if self.pairs:
_f = merge_casedata_ext(self._calc_data_cplx,
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig,
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:] += pair_difference_ext(_f, self.pairidx)
for i in range(self.parms - 1):
_g = merge_casedata_ext(self._gradients_cplx[:, i],
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_g -= merge_casedata_ext(self._gradients_lig[:, i],
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._g[:, i] += pair_difference_ext(_g, self.pairidx)
else:
self._f[:] = merge_casedata_ext(self._calc_data_cplx,
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:] -= merge_casedata_ext(self._calc_data,
self.ind_case, self.w,
self.ind_rec)
self._f[:] -= merge_casedata_ext(self._calc_data_lig,
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
for i in range(self.parms):
self._g[:,
i] = merge_casedata_ext(self._gradients_cplx[:, i],
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._g[:,
i] -= merge_casedata_ext(self._gradients[:, i],
self.ind_case, self.w,
self.ind_rec)
self._g[:,
i] -= merge_casedata_ext(self._gradients_lig[:, i],
self.ind_case_lig,
self.w_lig,
self.ind_case_lig)
self._f[:] += x[-1]
self._g[:, -1] = 1
else:
if self.pairs:
### Complex and Ligand contributions
_f = merge_casedata_ext(self._calc_data_cplx,
self.ind_case_cplx, self.w_cplx,
self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig,
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:] += pair_difference_ext(_f, self.pairidx)
else:
### Receptor contributions
self._f[:] = merge_casedata_ext(self._calc_data_cplx,
self.ind_case_cplx,
self.w_cplx,
self.ind_case_cplx)
self._f[:] -= merge_casedata_ext(self._calc_data,
self.ind_case, self.w,
self.ind_rec)
self._f[:] -= merge_casedata_ext(self._calc_data_lig,
self.ind_case_lig, self.w_lig,
self.ind_case_lig)
self._f[:] += x[-1]
def _f_process(self, x):
__doc___= """
returns the squared sum of residuals
objective function is the free energy
"""
self.gist_functional(x)
self._f[:] = 0.
if self.anal_grad:
self._g[:] = 0.
if self.pairs:
_f = merge_casedata_ext(self._calc_data_cplx[:,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,0] += pair_difference_ext(_f, self.pairidx)
_f = merge_casedata_ext(self._calc_data_cplx[:,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,1] += pair_difference_ext(_f, self.pairidx)
for i in range(self.parms-1):
_g = merge_casedata_ext(self._gradients_cplx[:,i,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_g -= merge_casedata_ext(self._gradients_lig[:, i,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._g[:,i,0] += pair_difference_ext(_g, self.pairidx)
_g = merge_casedata_ext(self._gradients_cplx[:,i,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_g -= merge_casedata_ext(self._gradients_lig[:, i,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._g[:,i,1] += pair_difference_ext(_g, self.pairidx)
else:
self._f[:,0] = merge_casedata_ext(self._calc_data_cplx[:,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._f[:,0] -= merge_casedata_ext(self._calc_data[:,0], self.ind_case, self.w, self.ind_rec)
self._f[:,0] -= merge_casedata_ext(self._calc_data_lig[:,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,1] = merge_casedata_ext(self._calc_data_cplx[:,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._f[:,1] -= merge_casedata_ext(self._calc_data[:,1], self.ind_case, self.w, self.ind_rec)
self._f[:,1] -= merge_casedata_ext(self._calc_data_lig[:,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
for i in range(self.parms+1):
self._g[:,i,0] = merge_casedata_ext(self._gradients_cplx[:,i,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._g[:,i,0] -= merge_casedata_ext(self._gradients[:,i,0], self.ind_case, self.w, self.ind_rec)
self._g[:,i,0] -= merge_casedata_ext(self._gradients_lig[:, i,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._g[:,i,1] = merge_casedata_ext(self._gradients_cplx[:,i,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._g[:,i,1] -= merge_casedata_ext(self._gradients[:,i,1], self.ind_case, self.w, self.ind_rec)
self._g[:,i,1] -= merge_casedata_ext(self._gradients_lig[:, i,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,0] += x[-2]
self._f[:,1] += x[-1]
self._g[:,-2,0] = 1
self._g[:,-2,1] = 0
self._g[:,-1,0] = 0
self._g[:,-1,1] = 1
else:
if self.pairs:
_f = merge_casedata_ext(self._calc_data_cplx[:,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,0] += pair_difference_ext(_f, self.pairidx)
_f = merge_casedata_ext(self._calc_data_cplx[:,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
_f -= merge_casedata_ext(self._calc_data_lig[:,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,1] += pair_difference_ext(_f, self.pairidx)
else:
self._f[:,0] = merge_casedata_ext(self._calc_data_cplx[:,0], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._f[:,0] -= merge_casedata_ext(self._calc_data[:,0], self.ind_case, self.w, self.ind_rec)
self._f[:,0] -= merge_casedata_ext(self._calc_data_lig[:,0], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,1] = merge_casedata_ext(self._calc_data_cplx[:,1], self.ind_case_cplx, self.w_cplx, self.ind_case_cplx)
self._f[:,1] -= merge_casedata_ext(self._calc_data[:,1], self.ind_case, self.w, self.ind_rec)
self._f[:,1] -= merge_casedata_ext(self._calc_data_lig[:,1], self.ind_case_lig, self.w_lig, self.ind_case_lig)
self._f[:,0] += x[-2]
self._f[:,1] += x[-1]