search

MarcLavielle / Rsmlx

Other
4 stars 2 forks source link

My data sets won't run via pkbuild #5

Open bfreed3 opened 3 years ago

bfreed3 commented 3 years ago

I'm currently running the following set up: Mac OSX 10.15.7, R 4.0.3, Monolix 2020R1 (also have 2019R2 installed), mlxR 4.1.7, lixoftConnectors 2020.1, Rsmlx 3.0.0 (dev version).

I'm trying to identify structural model with 2 particular data sets using pkbuild, but it always returns the error "Error in log(y) : non-numeric argument to mathematical function"

These same data sets run fine manually in Monolix with an oral1_VCl model, full diagonal, and combined1. Additionally the Rsmlx20_demos demo_pkbuild.R and remifentanil.R also both run fine for me.

I've attached my files as a zip for you to look at. Is there something wrong with my .csv data set format, because it only seems to fail like this when I use my own real data files.
Rsmlx_test 20210225.zip

Thanks, Burgess

MarcLavielle commented 3 years ago

pkbuild doesn't find any information about the absorption phase in the data... which looks much more like data from IV bolus are you sure it's oral administration ?

Marc

Le ven. 26 févr. 2021 à 17:05, bfreed3 notifications@github.com a écrit :

I'm currently running the following set up: Mac OSX 10.15.7, R 4.0.3, Monolix 2020R1 (also have 2019R2 installed), mlxR 4.1.7, lixoftConnectors 2020.1, Rsmlx 3.0.0 (dev version).

I'm trying to identify structural model with 2 particular data sets using pkbuild, but it always returns the error "Error in log(y) : non-numeric argument to mathematical function"

These same data sets run fine manually in Monolix with an oral1_VCl model, full diagonal, and combined1. Additionally the Rsmlx20_demos demo_pkbuild.R and remifentanil.R also both run fine for me.

I've attached my files as a zip for you to look at. Is there something wrong with my .csv data set format, because it only seems to fail like this when I use my own real data files. Rsmlx_test 20210225.zip https://github.com/MarcLavielle/Rsmlx/files/6050845/Rsmlx_test.20210225.zip

Thanks, Burgess

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/MarcLavielle/Rsmlx/issues/5, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACO7FOXN6TGKJKK4XNI5T73TA7BGDANCNFSM4YIWBX5A .

-- Marc Lavielle Research Director at Inria Head of the Xpop team

address:

CMAP, École polytechnique91128 Palaiseau Cedex

France Tel: (33) 1 69 33 46 00 (33) 6 37 31 93 05

Marc.Lavielle@inria.fr http://www.cmap.polytechnique.fr/~lavielle

"""""""""""""""""""""""""""""""""""""""""""""""""""""""""

bfreed3 commented 3 years ago

I would have to think that this has to do with how it sequences through the models w/ pkbuild in R specifically. I gather if you say the admin is "oral" and there's nothing in the absorption phase to evaluate, it can't decipher between oral1 or oral0 and it just crashes? With regard to the administration, it is mouse PK data and was oral gavage of solution in both instances.