def test_des(testdata):
code = """$PROBLEM
$INPUT ID TIME DV AMT
$DATA data.csv IGNORE=@
$SUBROUTINES ADVAN13 TOL=6
$MODEL COMP=(DOSE) COMP=(CENTRAL) COMP=(PERIPH) COMP=(LIVER)
$PK
MAT = THETA(1) * EXP(ETA(1))
CL = THETA(2) * EXP(ETA(2))
VC = THETA(3) * EXP(ETA(3))
Q = THETA(4) * EXP(ETA(4))
VP = THETA(5) * EXP(ETA(5))
VM = THETA(6) * EXP(ETA(6))
KM = THETA(7) * EXP(ETA(7))
KLO = THETA(8) * EXP(ETA(8))
VL = THETA(9) * EXP(ETA(9))
KA = 1/MAT
$DES
DADT(1) = -(KA * A(1))
DADT(2) = KA * A(1) + Q/VP * A(3) - (Q/VC * A(2) + A(2)/VC * VM/(KM + A(2)/VC))
DADT(3) = Q/VC * A(2) - Q/VP * A(3)
DADT(4) = A(2)/VC * VM/(KM + A(2)/VC) - KLO * A(4)
$ERROR
CONC = A(2)/VC
Y = CONC + EPS(1)
$ESTIMATION METHOD=COND INTER
$COVARIANCE PRINT=E
$THETA (0, 1, Inf) ; POP_MAT
$THETA (0, 1, Inf) ; POP_CL
$THETA (0, 1, Inf) ; POP_VC
$THETA (0, 1, Inf) ; POP_Q
$THETA (0, 1, Inf) ; POP_VP
$THETA (0, 1, Inf) ; POP_VM
$THETA (0, 1, Inf) ; POP_KM
$THETA (0, 1, Inf) ; POP_KLO
$THETA (0, 1, Inf) ; POP_VL
$OMEGA 0.1; IIV_MAT
$OMEGA 0.1; IIV_CL
$OMEGA 0.1; IIV_VC
$OMEGA 0.1; IIV_Q
$OMEGA 0.1; IIV_VP
$OMEGA 0.1; IIV_VM
$OMEGA 0.1; IIV_KM
$OMEGA 0.1; IIV_KLO
$OMEGA 0.1; IIV_VL
$SIGMA 0.1; RUV_ADD
"""
pheno = Model(testdata / 'nonmem' / 'pheno.mod')
model = Model(StringIO(code))
model.dataset = pheno.dataset
cs = model.statements.ode_system.to_compartmental_system()
assert len(cs) == 5
def test_get_params(testdata):
model_frem = Model(testdata / 'nonmem' / 'frem' / 'pheno' / 'model_4.mod')
rvs, _ = model_frem.random_variables.distributions(
level=VariabilityLevel.IIV)[-1]
npars = 2
param_names = get_params(model_frem, rvs, npars)
assert param_names == ['CL', 'V']
model_multiple_etas = re.sub(
r'(V=TVV\*EXP\(ETA\(2\)\))',
r'\1*EXP(ETA(3))',
str(model_frem),
)
model = Model(StringIO(model_multiple_etas))
model.dataset = model_frem.dataset
rvs, _ = model.random_variables.distributions(
level=VariabilityLevel.IIV)[-1]
npars = 3
param_names = get_params(model, rvs, npars)
assert param_names == ['CL', 'V(1)', 'V(2)']
model_separate_declare = re.sub(
r'(V=TVV\*EXP\(ETA\(2\)\))',
'ETA2=ETA(2)\n V=TVV*EXP(ETA2)',
str(model_frem),
)
model = Model(StringIO(model_separate_declare))
model.dataset = model_frem.dataset
rvs, _ = model.random_variables.distributions(
level=VariabilityLevel.IIV)[-1]
npars = 2
param_names = get_params(model, rvs, npars)
print(param_names)
assert param_names == ['CL', 'V']
def test_add_covariate_effect_nan(pheno_path):
model = Model(pheno_path)
data = model.dataset
new_col = [np.nan] * 10 + ([1.0] * (len(data.index) - 10))
data['new_col'] = new_col
model.dataset = data
add_covariate_effect(model, 'CL', 'new_col', 'cat')
model.update_source(nofiles=True)
assert not re.search('NaN', str(model))
assert re.search(r'NEW_COL\.EQ\.-99', str(model))
def test_check_covariates(testdata):
model = Model(testdata / 'nonmem' / 'pheno_real.mod')
newcov = check_covariates(model, ['WGT', 'APGR'])
assert newcov == ['WGT', 'APGR']
newcov = check_covariates(model, ['APGR', 'WGT'])
assert newcov == ['APGR', 'WGT']
data = model.dataset
data['NEW'] = data['WGT']
model.dataset = data
with pytest.warns(UserWarning):
newcov = check_covariates(model, ['APGR', 'WGT', 'NEW'])
assert newcov == ['APGR', 'WGT']
with pytest.warns(UserWarning):
newcov = check_covariates(model, ['NEW', 'APGR', 'WGT'])
assert newcov == ['NEW', 'APGR']
def test_check_covariates(testdata):
model = Model(testdata / 'nonmem' / 'pheno_real.mod')
newcov = check_covariates(model, ['FA1', 'FA2'])
assert newcov == []
newcov = check_covariates(model, ['WGT', 'APGR'])
assert newcov == ['WGT', 'APGR']
newcov = check_covariates(model, ['APGR', 'WGT'])
assert newcov == ['APGR', 'WGT']
data = model.dataset
data['NEW'] = data['WGT']
model.dataset = data
newcov = check_covariates(model, ['APGR', 'WGT', 'NEW'])
assert newcov == ['APGR', 'WGT']
newcov = check_covariates(model, ['NEW', 'APGR', 'WGT'])
assert newcov == ['NEW', 'APGR']
def test_power_on_ruv(testdata, epsilons, err_ref, theta_ref):
with Patcher(additional_skip_names=['pkgutil']) as patcher:
fs = patcher.fs
fs.add_real_file(testdata / 'nonmem/pheno_real.mod',
target_path='run1.mod')
fs.add_real_file(testdata / 'nonmem/pheno_real.phi',
target_path='run1.phi')
fs.add_real_file(testdata / 'nonmem/pheno_real.ext',
target_path='run1.ext')
fs.add_real_file(testdata / 'nonmem/pheno.dta',
target_path='pheno.dta')
model_pheno = Model('run1.mod')
model_more_eps = re.sub(
r'( 0.031128 ; IVV\n)',
'$SIGMA 0.1\n$SIGMA 0.1',
str(model_pheno),
)
model_more_eps = re.sub(
r'IPRED=F\nIRES=DV-IPRED',
r'IPRED=F+EPS(2)\nIRES=DV-IPRED+EPS(3)',
model_more_eps,
)
model = Model(StringIO(model_more_eps))
model.dataset = model_pheno.dataset
power_on_ruv(model, epsilons)
model.update_source()
rec_err = str(model.control_stream.get_records('ERROR')[0])
assert rec_err == f'$ERROR\n' f'W=F\n' f'{err_ref}\n' f'IWRES=IRES/W\n\n'
rec_theta = ''.join(
str(rec) for rec in model.control_stream.get_records('THETA'))
assert (rec_theta == f'$THETA (0,0.00469307) ; PTVCL\n'
f'$THETA (0,1.00916) ; PTVV\n'
f'$THETA (-.99,.1)\n'
f'{theta_ref}\n')
def psn_frem_results(path,
force_posdef_covmatrix=False,
force_posdef_samples=500,
method=None):
"""Create frem results from a PsN FREM run
:param path: Path to PsN frem run directory
:return: A :class:`FREMResults` object
"""
path = Path(path)
model_4_path = path / 'final_models' / 'model_4.mod'
if not model_4_path.is_file():
raise IOError(f'Could not find FREM model 4: {str(model_4_path)}')
model_4 = Model(model_4_path)
if model_4.modelfit_results is None:
raise ValueError('Model 4 has no results')
cov_model = None
if method == 'cov_sampling':
try:
model_4.modelfit_results.covariance_matrix
except Exception:
model_4b_path = path / 'final_models' / 'model_4b.mod'
try:
model_4b = Model(model_4b_path)
except FileNotFoundError:
pass
else:
cov_model = model_4b
with open(path / 'covariates_summary.csv') as covsum:
covsum.readline()
raw_cov_list = covsum.readline()
all_covariates = raw_cov_list[1:].rstrip().split(',')
# FIXME: Not introducing yaml parser in pharmpy just yet. Options should be collected
# differently. Perhaps using json
logtransformed_covariates = []
with open(path / 'meta.yaml') as meta:
for row in meta:
row = row.strip()
if row.startswith('rescale: 1'):
rescale = True
elif row.startswith('rescale: 0'):
rescale = False
if row.startswith("log: ''"):
logtransformed_covariates = []
elif row.startswith('log: '):
logtransformed_covariates = row[5:].split(',')
# add log transformed columns for the -log option. Should be done when creating dataset
df = model_4.dataset
if logtransformed_covariates:
for lncov in logtransformed_covariates:
df[f'LN{lncov}'] = np.log(df[lncov])
model_4.dataset = df
nunique = model_4.dataset.pharmpy.baselines[all_covariates].nunique()
continuous = list(nunique.index[nunique != 2])
categorical = list(nunique.index[nunique == 2])
intmod_names = [
'model_1.mod', 'model_2.mod', 'model_3.mod', 'model_3b.mod'
]
intmods = []
for m in intmod_names:
intmod_path = path / 'm1' / m
if intmod_path.is_file():
intmod = Model(intmod_path)
intmods.append(intmod)
res = calculate_results(
model_4,
continuous,
categorical,
method=method,
force_posdef_covmatrix=force_posdef_covmatrix,
force_posdef_samples=force_posdef_samples,
cov_model=cov_model,
rescale=rescale,
intermediate_models=intmods,
)
return res
