nanhung
commented
2 years ago Hi Brad,
I tested the code and found that the issue is because of the setting of the time point. The given time points are times <- seq(0, 48, 0.5). But, the concentration outputs are 0 at the beginning. In addition, there is no obvious uncertainty output. The issue can be solved by resetting the time points like times <- seq(10, 48, 0.5). I also noticed that only fub in the testing parameters have impacts on CVen, you can have more testing by using the code I provided as follow,
devtools::install_github("nanhung/RMCSim")
library(RMCSim)
library(pksensi)
install_mcsim()
model <- "dmf.model"
makemcsim(model = model, dir = "MCSim") # The model file is placed in the subfolder named "MCSim" in the working directory
q <- "qunif"
condition <- c("InhMag = 10", "InhDur = 7", "InhStart = 2", "InhPeriod = 48",
"PCAir = 1", "log_PCFat = 1", "Frac = 0.8", "kGut = 0.3",
"Kp_sc_vs = 0.2", "fSA_exposed = 0.7", "Ke = 1",
"Michaelis = 0", "Vmax = 10", "Km = 1", "CLH = 10")
params <- c("fub", "Ke", "CLH")
outputs <- c("CVen")
dist <- rep("Uniform", 3)
fub <- 0.5
Ke <- 1
CLH <- 1
q.arg <- list(list(min = fub * 0.8, max = fub * 1.2),
list(min = Ke * 0.8, max = Ke * 1.2),
list(min = CLH * 0.8, max = CLH * 1.2))
pk.times <- seq(0, 48, 0.5)
# PK plot -----------------------------------------------------------------
set.seed(1111)
out <- solve_mcsim(mName = model, params = params, vars = outputs,
monte_carlo = 1000, dist = dist, q.arg = q.arg,
time = pk.times, condition = condition)
# Check PK plot
pksim(out, vars = "CVen", xlab = "Time", ylab = "Conc.", legend = FALSE)
# Sensitivity analysis ----------------------------------------------------
set.seed(1234)
x <- rfast99(params = params, n = 512, q = q, q.arg = q.arg, replicate = 5)
# Reproduce error message
times1 <- seq(0, 48, 0.5)
out1 <- solve_mcsim(x, mName = model, params = params, time = times1, vars = outputs, condition = condition)
## Error in quantile.default(newX[, i], ...) :
## missing values and NaN's not allowed if 'na.rm' is FALSE
# Change time points. Started from 10 hr
times2 <- seq(10, 48, 0.5) # Set from 10 to 48 hr since the initial time points have no uncertainty output
out2 <- solve_mcsim(x, mName = model, params = params, time = times2, vars = outputs, condition = condition)
check(out2)
plot(out2) # Only fub have impact on CVen
# Additional check --------------------------------------------------------
# Set the lower fub (0.5 -> 0.1)
fub <- 0.1
q.arg <- list(list(min = fub * 0.8, max = fub * 1.2),
list(min = Ke * 0.8, max = Ke * 1.2),
list(min = CLH * 0.8, max = CLH * 1.2))
q.arg # fub: U(0.08, 0.12)
set.seed(1111)
out3 <- solve_mcsim(mName = model, params = params, vars = outputs,
monte_carlo = 1000, dist = dist, q.arg = q.arg,
time = times1, condition = condition)
pksim(out3)
# Set Ke (1 -> 10)
Ke <- 10
q.arg <- list(list(min = fub * 0.8, max = fub * 1.2),
list(min = Ke * 0.8, max = Ke * 1.2),
list(min = CLH * 0.8, max = CLH * 1.2))
q.arg # Ke: U(8, 12)
set.seed(1111)
out4 <- solve_mcsim(mName = model, params = params, vars = outputs,
monte_carlo = 1000, dist = dist, q.arg = q.arg,
time = times1, condition = condition)
pksim(out4)
# Does not have change on the output
# Set CLH (1 -> 100)
CLH <- 100
q.arg <- list(list(min = fub * 0.8, max = fub * 1.2),
list(min = Ke * 0.8, max = Ke * 1.2),
list(min = CLH * 0.8, max = CLH * 1.2))
q.arg # CLH: U(80, 120)
set.seed(1111)
out5 <- solve_mcsim(mName = model, params = params, vars = outputs,
monte_carlo = 1000, dist = dist, q.arg = q.arg,
time = times1, condition = condition)
pksim(out5)
# Does not have change on the output
breisfeld
Hi Nan-Hung,
I am trying to run
pksension a pbpk model that I've adapted from one created by Frederic.The following is my job file:
Running the file in RStudio gives the following:
Attached is the model file.
I didn't see anything obvious in the code for
run_mcsim, but I am not an R expert. There is no traceback, so I don't know where to begin.Thanks.
-Brad
dmf.model.txt