xpose.nlmixr does not capture all of the data

[mattfidler]

There seems to be a problem importing data into xpose; Some of the data isn't captured.

See:

UUPharmacometrics/xpose#154

This can be useful in:

https://github.com/nlmixrdevelopment/nlmixr-examples/blob/master/case-study-warfarin/warfarin_PKPD_joint.R

@kestrel99 do you have any idea on this? It shouldn't be hard to be implement?

[bguiastr]

As part of the upcoming initiative to make xpose S3 compatible, I will also try to contribute to xpose.nlmixr and bring it up to speed.

[mattfidler]

Thank you @guiastrennec; Its much appreciated.

[DedeLLL]

Hi Dr.@mattfidler I am not sure if we are talking about the same problem. I was using the latest nlmixr 1.0.0.8 on github and wanted to use the function xpose_data_nlmixr() to convert the data. However I got the same error every time.
Error in data.frame(data_a, full.dat) : arguments imply differing number of rows: 6960, 6947
I also tried the old version of nlmixr 1.0.07 from Cran and the function xpose_data_nlmixr() worked smoothly. The package xpose.nlmixr I was using is version 0.1.3 downloaded from github.
Thanks for your time! :)

[kestrel99]

Thanks @guiastrennec! @DedeLLL, I'll look into your issue as well.

[DedeLLL]

Thanks so much! @kestrel99 :)

[mattfidler]

Hi @DedeLLL it seems like a different issue. Was there any warnings after fitting the data, like IDs that were dropped?

[DedeLLL]

Hi @mattfidler @kestrel99 Thanks for your time, I actually used the example from the nlmixr package help document.

> d <- Oral_1CPT
  d <- d[,names(d) != "SS"]
  f <- function(){
    ini({
        lCl <- 1.6     
        lVc <- log(90)   
        lKA <- 0.1     
        prop.err <- c(0, 0.2, 1)
        eta.Cl ~ 0.1 
        eta.Vc ~ 0.1 
        eta.KA ~ 0.1 
    })
    model({
        Cl <- exp(lCl + eta.Cl)
        Vc = exp(lVc + eta.Vc)
        KA <- exp(lKA + eta.KA)
        kel <- Cl / Vc;
        d/dt(depot)    = -KA*depot;
        d/dt(centr)  =  KA*depot-kel*centr;
        cp = centr / Vc;
        cp ~ prop(prop.err)
    })
}
fit.s <-  nlmixr(f,d,est="saem",control=saemControl(n.burn=50,n.em=100,print=50))

The result I got was

Warning message:
In (function (uif, data, est = NULL, control = list(), ..., sum.prod = FALSE,  :
  Bounds are ignored in SAEM

This was the only warning I got.
Then I used

fit.s <- addCwres(fit.s)
a<-xpose_data_nlmixr(fit.s)

I got the error

Error in data.frame(data_a, full.dat) : 
  arguments imply differing number of rows: 6960, 6947`

[DedeLLL]

Also I tried another dataset from the help document of nlmixr.
fit.s <- nlmixr(f,Bolus_2CPT,est="saem",control=saemControl(n.burn=50,n.em=100))
f is the same model.
a<-xpose_data_nlmixr(fit.s)
I got another error
Error in suppessWarnings(do.call(RxODE:::rxSolve.default, c(list(object = .rx, : could not find function "suppessWarnings"

[mattfidler]

Hi @DedeLLL ,

I believe your issues have been addressed. Please update RxODE, nlmixr and xpose.nlmixr.

Here is my example and the results:

library(nlmixr)
library(xpose.nlmixr)
#> Loading required package: xpose
#> Loading required package: ggplot2
#> 
#> Attaching package: 'xpose'
#> The following object is masked from 'package:nlmixr':
#> 
#>     vpc
#> The following object is masked from 'package:stats':
#> 
#>     filter
#> 
#> Attaching package: 'xpose.nlmixr'
#> The following object is masked from 'package:nlmixr':
#> 
#>     vpc

d <- Oral_1CPT;
d <- d[d$ID %in% unique(d$ID)[1:12], ]
# d <- d[,names(d) != "SS"] ## SS supported in RxODE/nlmixr now
f <- function(){
    ini({
        lCl <- 1.6
        lVc <- log(90)
        lKA <- 0.1
        prop.err <- c(0, 0.2, 1)
        eta.Cl ~ 0.1
        eta.Vc ~ 0.1
        eta.KA ~ 0.1
    })
    model({
        Cl <- exp(lCl + eta.Cl)
        Vc = exp(lVc + eta.Vc)
        KA <- exp(lKA + eta.KA)
        kel <- Cl / Vc;
        d/dt(depot)    = -KA*depot;
        d/dt(centr)  =  KA*depot-kel*centr;
        cp = centr / Vc;
        cp ~ prop(prop.err)
    })
}

fit.s <-  nlmixr(f,d,est="saem",control=saemControl(n.burn=50,n.em=100,print=50))
#> Warning in (function (uif, data, est = NULL, control = list(), ...,
#> sum.prod = FALSE, : Bounds are ignored in SAEM
#> Compiling RxODE equations...
#> done.
#> 1:    1.8670   4.5645   0.0235   0.0950   0.0964   0.0950   0.5929
#> 50:    1.4771   4.1704   0.0226   0.0873   0.1208   0.1207   0.2014
#> 100:    1.4764   4.1758   0.0076   0.0906   0.1233   0.1194   0.2015
#> 150:    1.4737   4.1723   0.0045   0.0890   0.1263   0.1182   0.2009
#> Calculating residuals/tables
#> done.

xpdb <- xpose_data_nlmixr(fit.s)
#> Calculating residuals/tables
#> done.
#> Warning in xpose_data_nlmixr(fit.s): Added CWRES to fit (using
#> nlmixr::addCwres)

dv_vs_pred(xpdb)

^{Created on 2019-04-07 by the reprex package (v0.2.1)}

[DedeLLL]

It works! Thanks sooooooo much @mattfidler for your precious time and effort. By the way, I updated all the new packages for macOSX and it also worked really well. (The previous versions could not work on my mac although I checked my clang and gfortran to be the latest version) That is a real bonus!

[mattfidler]

Great! I'm glad to hear it.