def test_reweight_initialize():
"""Tests the initialization and method parameters of a
MaxEnt reweighting object"""
test_dict = {'temp': 350, 'do_reweight': False, 'test_param': 'foo'}
expected_dict = {
'do_reweight': False,
'do_params': True,
'do_mcmin': False,
'do_mcsampl': False,
'mc_refvar': 0.03,
'mc_equilsteps': -1,
'bv_bc': 0.35,
'bv_bh': 2.00,
'bv_bcrange': 1.5,
'bv_bhrange': 16.0,
'tolerance': 10**-10,
'maxiters': 10**6,
'param_maxiters': 10**2,
'stepfactor': 10**-5,
'stepfactor_scaling': 1.005,
'param_stepfactor': 10**-1,
'temp': 350,
'random_initial': False,
'kT': 2.9101093,
'test_param': 'foo'
}
test_obj = reweighting.MaxEnt(**test_dict)
assert test_obj.methodparams == expected_dict
def test_optimize_parameters_gradient():
"""Test that beta_c and beta_h parameters are optimized correctly
with the gradient-descent protocol"""
test_prefix = 'HDXer/tests/data/tmp_test_output/tmp_'
test_folders = ['HDXer/tests/data/reweighting_1']
test_kint_file = os.path.join('HDXer/tests/data/reweighting_1',
'intrinsic_rates.dat')
# Residue-based, calculated with Bc = 0.25, Bh = 5.25
test_exp_file = os.path.join('HDXer/tests/data/reweighting_3',
'experimental_data_new_params.dat')
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_times = np.array([0.5, 5.0, 60.0])
test_param_dict = {
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_obj = reweighting.MaxEnt(param_maxiters=10000)
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
# Set initial values of lnpi etc. using data read in
test_obj.update_lnpi_and_weights()
test_obj.update_dfracs_and_mse()
# Calculated with Bc = 0.25, Bh = 5.25
expected_bc, expected_bh = 0.25, 5.25
test_obj.optimize_parameters_gradient()
out_bc = test_obj.methodparams['bv_bc']
out_bh = test_obj.methodparams['bv_bh']
out_mse = test_obj.runvalues['curr_MSE']
assert np.isclose(out_bc, expected_bc)
assert np.isclose(out_bh, expected_bh)
assert np.isclose(out_mse, 0.0)
def test_run_full_reweight_4():
"""Tests that output files are created correctly using file comparisons.
Full reweight of experimental data"""
test_obj = reweighting.MaxEnt(do_reweight=True)
test_prefix = 'HDXer/tests/data/tmp_test_output/test_run_full_reweight_'
reference_prefix = 'HDXer/tests/data/reweighting_4/reference_run_full_reweight_'
test_folders = ['HDXer/tests/data/reweighting_4']
test_kint_file = os.path.join('HDXer/tests/data/reweighting_4',
'intrinsic_rates.dat')
test_exp_file = os.path.join('HDXer/tests/data/reweighting_4',
'experimental_data.dat')
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_times = np.array([5.0, 30.0])
test_restart = 50
test_param_dict = {
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix,
'stepfactor': 10**-5,
'restart_interval': test_restart
}
test_obj.run(gamma=10**-2, **test_param_dict)
test_suffixes = [
'final_segment_fractions.dat', 'final_weights.dat',
'initial_params.dat', 'per_iteration_output.dat',
'per_restart_output.dat', 'work.dat'
]
test_files = [test_prefix + f for f in test_suffixes]
expected_files = [reference_prefix + f for f in test_suffixes]
# compare file contents with filecmp
for out_file, expected_file in zip(test_files, expected_files):
assert filecmp.cmp(out_file, expected_file, shallow=False)
def test_reweight_data_io_1():
"""Test that runparams are assigned correctly and
data files are read in correctly.
Test 1 checks contacts & H-bond inputs"""
expected_nframes = 10
expected_hbonds = np.array([[1, 1, 1, 0, 0, 2, 0, 0, 0, 0],
[1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
[2, 2, 2, 2, 2, 1, 1, 1, 1, 1],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0, 0, 1, 0],
[1, 0, 2, 0, 1, 0, 2, 0, 1, 0]])
expected_contacts = np.array([[10, 10, 10, 5, 5, 20, 5, 5, 5, 5],
[10, 10, 10, 10, 10, 10, 10, 10, 10, 10],
[20, 20, 20, 20, 20, 10, 10, 10, 10, 10],
[10, 10, 10, 5, 5, 20, 5, 5, 5, 5],
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[4, 4, 22, 4, 11, 4, 23, 3, 10, 5]])
expected_kint = np.array([10.0, 100.0, 1000.0, 0.1, 268.21, 8.5])
test_times = np.array([0.5, 5.0, 60.0])
expected_minuskt = -expected_kint[:, np.newaxis] * test_times
expected_expt = np.array([[0.72823031, 0.83269571, 0.97653617],
[0.75439649, 0.79088392, 0.97067021],
[0.99994366, 1.00000000, 1.00000000]])
test_prefix = 'HDXer/tests/data/tmp_test_output/tmp_'
test_folders = ['HDXer/tests/data/reweighting_1']
test_kint_file = os.path.join('HDXer/tests/data/reweighting_1',
'intrinsic_rates.dat')
test_exp_file = os.path.join('HDXer/tests/data/reweighting_1',
'experimental_data.dat')
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_param_dict = {
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_obj = reweighting.MaxEnt()
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
assert np.array_equal(test_obj.runvalues['contacts'], expected_contacts)
assert np.array_equal(test_obj.runvalues['hbonds'], expected_hbonds)
assert test_obj.runvalues['nframes'] == expected_nframes
# Broadcast expected_minuskt to 3 dims for segments * residues * times
assert np.array_equal(
test_obj.runvalues['minuskt'],
np.repeat(expected_minuskt[np.newaxis, :, :],
len(expected_expt),
axis=0))
# Internally the exp_dfrac_filtered arrays have 0 for residues not in the segment.
# So, test for closeness, not identity
test_expt_normalised = np.sum(test_obj.runvalues['exp_dfrac_filtered'], axis=1) \
/ np.sum(test_obj.runvalues['segfilters'], axis=1)
assert np.allclose(test_expt_normalised, expected_expt)
def test_update_lnpi_2():
"""Test that weights are calculated correctly with provided lambdas"""
test_prefix = 'HDXer/tests/data/tmp_test_output/tmp_'
test_folders = ['HDXer/tests/data/reweighting_1']
test_kint_file = os.path.join('HDXer/tests/data/reweighting_1',
'intrinsic_rates.dat')
test_exp_file = os.path.join('HDXer/tests/data/reweighting_1',
'experimental_data.dat')
test_times = np.array([0.5, 5.0, 60.0])
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_param_dict = {
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_methodparam_dict = {'bv_bc': 0.25, 'bv_bh': 5.25}
test_obj = reweighting.MaxEnt(**test_methodparam_dict)
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
# Now apply lambdas and re-test weights are calculated correctly
test_lambdas = np.ones(6)
test_obj.runvalues['lambdas'] = test_lambdas
test_obj.update_lnpi_and_weights()
# Calculated with Bc = 0.25, Bh = 5.25
expected_lnpi_newbetas = np.array(
[[7.75, 7.75, 7.75, 1.25, 1.25, 15.5, 1.25, 1.25, 1.25, 1.25],
[7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75],
[15.5, 15.5, 15.5, 15.5, 15.5, 7.75, 7.75, 7.75, 7.75, 7.75],
[2.5, 2.5, 2.5, 1.25, 1.25, 5., 1.25, 1.25, 1.25, 1.25],
[0., 5.25, 0., 0., 0., 0., 0., 0., 5.25, 0.],
[6.25, 1., 16., 1., 8., 1., 16.25, 0.75, 7.75, 1.25]])
# Calculated manually as code below:
# biasfactor = np.sum(test_lambdas[:, np.newaxis] * expected_lnpi_newbetas, axis=0)
# expected_currweights = expected_iniweights * np.exp(biasfactor)
# expected_currweights = expected_currweights / np.sum(expected_currweights)
# expected_avelnpi_newbetas = np.sum(expected_currweights * expected_lnpi_newbetas, axis=1)
expected_currweights = np.array([
5.82876279e-05, 5.82876279e-05, 9.99879306e-01, 1.31749240e-10,
1.44480585e-07, 3.72620339e-06, 2.38208213e-07, 4.41969461e-14,
9.23633476e-09, 7.28684451e-14
])
expected_avelnpi_newbetas = np.array([
7.75002633e+00, 7.75000000e+00, 1.54999692e+01, 2.50000883e+00,
3.06058537e-04, 1.59985003e+01
])
expected_shape = (
3, 6, 3) # ave_lnpi is array of shape [n_segs, n_residues, n_times]
expected_avelnpi_newbetas = np.broadcast_to(
expected_avelnpi_newbetas[np.newaxis, :, np.newaxis], expected_shape)
assert np.allclose(test_obj.runvalues['lnpi'], expected_lnpi_newbetas)
assert test_obj.runvalues['ave_lnpi'].shape == expected_shape
assert np.allclose(test_obj.runvalues['ave_lnpi'],
expected_avelnpi_newbetas)
assert np.allclose(test_obj.runvalues['currweights'], expected_currweights)
assert np.isclose(np.sum(test_obj.runvalues['currweights']), 1.0)
# Case with -ve lambda, should still work
test_lambdas = np.array([0, 0, -0.1, 0.5, 0.00005, 2])
test_obj.runvalues['lambdas'] = test_lambdas
test_obj.update_lnpi_and_weights()
# Calculated manually as code below:
# biasfactor = np.sum(test_lambdas[:, np.newaxis] * expected_lnpi_newbetas, axis=0)
# expected_currweights = expected_iniweights * np.exp(biasfactor)
# expected_currweights = expected_currweights / np.sum(expected_currweights)
# expected_avelnpi_newbetas = np.sum(expected_currweights * expected_lnpi_newbetas, axis=1)
expected_currweights = np.array([
1.16557025e-09, 3.21040924e-14, 3.42989521e-01, 1.71795721e-14,
2.06602270e-08, 2.43160328e-13, 6.57010430e-01, 2.26174337e-14,
2.72069635e-08, 6.14805589e-14
])
expected_avelnpi_newbetas = np.array([
3.47943189e+00, 7.75000000e+00, 1.04081690e+01, 1.67873690e+00,
1.42836727e-07, 1.61642522e+01
])
expected_shape = (
3, 6, 3) # ave_lnpi is array of shape [n_segs, n_residues, n_times]
expected_avelnpi_newbetas = np.broadcast_to(
expected_avelnpi_newbetas[np.newaxis, :, np.newaxis], expected_shape)
assert np.allclose(test_obj.runvalues['lnpi'], expected_lnpi_newbetas)
assert test_obj.runvalues['ave_lnpi'].shape == expected_shape
assert np.allclose(test_obj.runvalues['ave_lnpi'],
expected_avelnpi_newbetas)
assert np.allclose(test_obj.runvalues['currweights'], expected_currweights)
assert np.isclose(np.sum(test_obj.runvalues['currweights']), 1.0)
def test_update_lnpi_1():
"""Test that iniweights, BV parameters are assigned correctly
and that weights are calculated correctly with lambda = 0"""
expected_iniweights = np.ones(10)
expected_currweights = np.ones(10) / 10
expected_lnpi = np.array(
[[5.5, 5.5, 5.5, 1.75, 1.75, 11., 1.75, 1.75, 1.75, 1.75],
[5.5, 5.5, 5.5, 5.5, 5.5, 5.5, 5.5, 5.5, 5.5, 5.5],
[11., 11., 11., 11., 11., 5.5, 5.5, 5.5, 5.5, 5.5],
[3.5, 3.5, 3.5, 1.75, 1.75, 7., 1.75, 1.75, 1.75, 1.75],
[0., 2., 0., 0., 0., 0., 0., 0., 2., 0.],
[3.4, 1.4, 11.7, 1.4, 5.85, 1.4, 12.05, 1.05, 5.5, 1.75]])
expected_avelnpi = np.array([3.8, 5.5, 8.25, 2.8, 0.4, 4.55])
expected_shape = (
3, 6, 3) # Shape of ave_lnpi array as [n_segs, n_residues, n_times]
expected_avelnpi = np.broadcast_to(
expected_avelnpi[np.newaxis, :, np.newaxis], expected_shape)
# Calculated with Bc = 0.25, Bh = 5.25
expected_lnpi_newbetas = np.array(
[[7.75, 7.75, 7.75, 1.25, 1.25, 15.5, 1.25, 1.25, 1.25, 1.25],
[7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75, 7.75],
[15.5, 15.5, 15.5, 15.5, 15.5, 7.75, 7.75, 7.75, 7.75, 7.75],
[2.5, 2.5, 2.5, 1.25, 1.25, 5., 1.25, 1.25, 1.25, 1.25],
[0., 5.25, 0., 0., 0., 0., 0., 0., 5.25, 0.],
[6.25, 1., 16., 1., 8., 1., 16.25, 0.75, 7.75, 1.25]])
expected_avelnpi_newbetas = np.array(
[4.625, 7.75, 11.625, 2., 1.05, 5.925])
expected_avelnpi_newbetas = np.broadcast_to(
expected_avelnpi_newbetas[np.newaxis, :, np.newaxis], expected_shape)
test_prefix = 'HDXer/tests/data/tmp_test_output/tmp_'
test_folders = ['HDXer/tests/data/reweighting_1']
test_kint_file = os.path.join('HDXer/tests/data/reweighting_1',
'intrinsic_rates.dat')
test_exp_file = os.path.join('HDXer/tests/data/reweighting_1',
'experimental_data.dat')
test_times = np.array([0.5, 5.0, 60.0])
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_param_dict = {
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_obj = reweighting.MaxEnt()
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
test_obj.update_lnpi_and_weights()
assert np.array_equal(test_obj.runvalues['iniweights'],
expected_iniweights)
assert np.allclose(test_obj.runvalues['lnpi'], expected_lnpi)
assert test_obj.runvalues['ave_lnpi'].shape == expected_shape
assert np.allclose(test_obj.runvalues['ave_lnpi'], expected_avelnpi)
test_obj.methodparams['bv_bc'] = 0.25
test_obj.methodparams['bv_bh'] = 5.25
test_obj.update_lnpi_and_weights()
assert np.allclose(test_obj.runvalues['lnpi'], expected_lnpi_newbetas)
assert np.allclose(test_obj.runvalues['ave_lnpi'],
expected_avelnpi_newbetas)
assert np.array_equal(test_obj.runvalues['currweights'],
expected_currweights)
def test_reweight_data_io_2():
"""Test that runparams are assigned correctly and
data files are read in correctly.
Test 2 subsamples contacts & H-bond inputs"""
# Do test of sub_start & sub_end
expected_nframes = 5
expected_hbonds = np.array([[1, 1, 1, 0, 0], [1, 1, 1, 1, 1],
[2, 2, 2, 2, 2], [0, 0, 0, 0, 0],
[0, 1, 0, 0, 0], [1, 0, 2, 0, 1]])
expected_contacts = np.array([[10, 10, 10, 5, 5], [10, 10, 10, 10, 10],
[20, 20, 20, 20, 20], [10, 10, 10, 5, 5],
[0, 0, 0, 0, 0], [4, 4, 22, 4, 11]])
test_prefix = 'HDXer/tests/data/tmp_test_output/tmp_'
test_folders = ['HDXer/tests/data/reweighting_1']
test_times = np.array([0.5, 5.0, 60.0])
test_kint_file = os.path.join('HDXer/tests/data/reweighting_1',
'intrinsic_rates.dat')
test_exp_file = os.path.join('HDXer/tests/data/reweighting_1',
'experimental_data.dat')
test_contacts_prefix = 'Contacts_chain_0_res_'
test_hbonds_prefix = 'Hbonds_chain_0_res_'
test_param_dict = {
'do_subsample': True,
'sub_start': 0,
'sub_end': 5,
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_obj = reweighting.MaxEnt()
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
# Check test of sub_start & sub_end
assert np.array_equal(test_obj.runvalues['contacts'], expected_contacts)
assert np.array_equal(test_obj.runvalues['hbonds'], expected_hbonds)
assert test_obj.runvalues['nframes'] == expected_nframes
# Do test of sub_interval
expected_nframes = 5
expected_hbonds = np.array([[1, 0, 2, 0, 0], [1, 1, 1, 1, 1],
[2, 2, 1, 1, 1], [0, 0, 0, 0, 0],
[1, 0, 0, 0, 0], [0, 0, 0, 0, 0]])
expected_contacts = np.array([[10, 5, 20, 5, 5], [10, 10, 10, 10, 10],
[20, 20, 10, 10, 10], [10, 5, 20, 5, 5],
[0, 0, 0, 0, 0], [4, 4, 4, 3, 5]])
test_param_dict = {
'do_subsample': True,
'sub_start': 1,
'sub_end': -1,
'sub_interval': 2,
'hbonds_prefix': test_hbonds_prefix,
'contacts_prefix': test_contacts_prefix,
'data_folders': test_folders,
'kint_file': test_kint_file,
'exp_file': test_exp_file,
'times': test_times,
'out_prefix': test_prefix
}
test_obj = reweighting.MaxEnt()
test_obj.set_run_params(10**-2, None, None, None, test_param_dict)
test_obj.setup_no_runobj(test_obj.runparams['data_folders'],
test_obj.runparams['kint_file'],
test_obj.runparams['exp_file'],
test_obj.runparams['times'])
# Check test of sub_interval
assert np.array_equal(test_obj.runvalues['contacts'], expected_contacts)
assert np.array_equal(test_obj.runvalues['hbonds'], expected_hbonds)
assert test_obj.runvalues['nframes'] == expected_nframes