sebastianueckert
commented
3 years ago I think there are two issues with these examples:
- assemblerr is case sensitive (i.e., q1 and Q1 are two different parameters)
- when providing the parameter name to the function, one needs to pay attention to the argument order or use named arguments so the following should work
pkmb2 <- pk_model() +
pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) +
pk_distribution_3cmp(prm_q1 = prm_no_var("q1")) +
pk_elimination_linear_nl() +
obs_proportional(conc~C["central"])
check(pkmb2)
render(pkmb2)
I did a couple of tests with the function prm_no_var, to get rid of some BSV on parameters. Depending in which the functions (pk_absorption_fo_transit or pk_distribution_3cmp) the function prm_no_var is embedded or depending on the parameters affected by the prm_no_var or if the parameter name is written with upper cases or lower cases, I get the NM code I expected or not
Trying the function prm_no_var on mat and vc (or written VC) I get the expected NM control file
pkmb <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp(prm_no_var("vc")) + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb) render(pkmb)
Trying with another parameter, q1
pkmb2 <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp(prm_no_var("q1")) + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb2) render(pkmb2)
I get the following message and nothing is done, there is still BSV on Q1
Warning message: Building block replaced x The building block [q1: no variability] has been replaced with [q1: log-normal]
When the parameter q1 is written with upper case, No warning but an additional parameter is created
pkmb3 <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp(prm_no_var("Q1")) + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb3) render(pkmb3)
$PK MDT = THETA(1) EXP(ETA(1)) MAT = THETA(2) Q1 = THETA(3) VP1 = THETA(4) EXP(ETA(2)) VP2 = THETA(5) EXP(ETA(3)) Q1 = THETA(6) EXP(ETA(4)) Q2 = THETA(7) EXP(ETA(5)) CLIN = THETA(8) EXP(ETA(6)) VMAX = THETA(9) EXP(ETA(7)) KM = THETA(10) EXP(ETA(8)) KTR = 3/MDT KA = 1/MAT (...) $THETA (0, 0.5, Inf) ; POP_MDT $THETA (0, 1, Inf) ; POP_MAT $THETA (0, 1, Inf) ; POP_Q1 $THETA (0, 5, Inf) ; POP_VP1 $THETA (0, 5, Inf) ; POP_VP2 $THETA (0, 25, Inf) ; POP_Q1 $THETA (0, 25, Inf) ; POP_Q2 $THETA (0, 50, Inf) ; POP_CLIN $THETA (0, 10, Inf) ; POP_VMAX $THETA (0, 0.5, Inf) ; POP_KM
Trying with the function not embedded in pk_distribution_3cmp() and upper case, No warning but an additional parameter is created as above
pkmb4 <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp() + prm_no_var("Q1") + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb4) render(pkmb4)
Trying with the function not embedded in pk_distribution_3cmp() and lower case, there NM code as expected
pkmb5 <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp() + prm_no_var("q1") + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb5) render(pkmb5)
Finally trying the function not embedded in pk_distribution_3cmp() and lower case with another parameter, clin, there is still BSV on CLIN
pkmb6 <- pk_model() + pk_absorption_fo_transit(transit_compartments = 3, prm_no_var("mat")) + pk_distribution_3cmp() + prm_no_var("clin") + pk_elimination_linear_mm() + obs_proportional(conc~C["central"]) check(pkmb6) render(pkmb6)