error in SCM - Githubissues

aargikar commented 4 years ago

Hello, I am new to NONMEM and PsN. I am trying to run scm and having a issue with it. I am getting the error below. If I run the same file on another PC, it seems to run without this error. I am using windows 10 and NONMEM 7.4.1. Can anyone suggest a solution for this issue? Please do not hesistate to contact me for additional information.

Calculating continuous covariate statistics
|........................................| ... done
Calculating categorical covariate statistics
|........................................| ... done
Starting scm forward search
Starting 8 NONMEM executions. 4 in parallel.
S:1 .. S:2 .. S:3 .. S:4 .. S:5 .. S:6 .. S:7 .. S:8 .. Waiting for all NONMEM runs to finish:
F:3 .. F:6 .. F:8 .. F:7 .. done
Argument " FAILED" isn't numeric in numeric gt (>) at C:\Perl64\site\lib\PsN_4_9_0/tool/scm.pm line 3696.
Adding AGE on CL state 5
Taking a step forward
Starting 7 NONMEM executions. 4 in parallel.
S:1 .. S:2 .. S:3 .. S:4 .. NMtran failed. There is no output for model 1. Contents of FMSG:

AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.

AN ERROR WAS FOUND ON LINE 119 AT THE APPROXIMATE POSITION NOTED:
$ETAS development\scm-model-dev-31\m1\CLAGE5.phi FILE=Model
X
27 NUMERIC VALUE IS FORMED INCORRECTLY (ERROR IN READ).

Not restarting this model.
S:5 .. NMtran failed. There is no output for model 2. Contents of FMSG:

AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.

AN ERROR WAS FOUND ON LINE 124 AT THE APPROXIMATE POSITION NOTED:
$ETAS development\scm-model-dev-31\m1\CLAGE5.phi FILE=Model
X
27 NUMERIC VALUE IS FORMED INCORRECTLY (ERROR IN READ).

rikardn commented 4 years ago

Which version of PsN do you use? Some of the scm-models are failing. Did your model have $ETAS before running scm?

It could be that you have a space in your path and that scm does not support that. Please try to run again from a path that does not contain space.

aargikar commented 4 years ago

Hi rikard, I am using PsN version 4.9. I do have ETAs in the model (before running scm). I don't have spaces in the path. My colleague and I verified through Perl Manager that all packages in Perl have been installed. I really appreciate your input on this issue as I am on a very tight timeline. Please let me know if you would like to see the .scm and .mod files.

rikardn commented 4 years ago

Ok, then I would guess that this is a bug.

The error from NM-TRAN points to an error in $ETAS. It could be something that PsN has created.

Can it run without $ETAS?
I would like to see your $ETAS from the input model

aargikar commented 4 years ago

I havent tried without ETAs. I will give it a try. Please see the select sections from control stream file below.

$PK
TVKA= THETA(2)
KA = TVKA * EXP(ETA(1))

TVCL = THETA(3)
CL = TVCL * EXP(ETA(2))

TVV2 = THETA(4)
V2 = TVV2 * EXP(ETA(3))

TVQ = THETA(5)
Q = TVQ * EXP(ETA(4))

TVV3 = THETA(6)
V3 = TVV3 * EXP(ETA(5))

TLAG1 = THETA(7) 
ALAG1=TLAG1

S2 = V2/1000; Scaler

$OMEGA 
 0.25 ; IIV KA 
 0.1 ; IIV CL
 0.5 ; IIV V2
 1.9 ; IIV Q
 1.1 ; IIV V3

rikardn commented 4 years ago

I don't see a $ETAS record in your snippet.

aargikar commented 4 years ago

The $OMEGA section are the primary estimates in the base model file (.mod file) . Are you referring to .scm file?

rikardn commented 4 years ago

The sections in a NONMEM model file (aka control stream) that starts with a $-sign are called records. The error message from your scm run comes from NONMEM, and it signals that something is wrong with the $ETA record. Do you have such a record in your input model, and in that case what does it look like.

aargikar commented 4 years ago

I don't have a $ETA record specifically in the control stream file in the input model. I have specified all ETAs under $OMEGA.

rikardn commented 4 years ago

This is because of two issues in PsN.

The $ETAs was added by PsN. This should only be done by default if you are using the linearize option, but it does it also for non-linear scm:s
I still think you have a space in your path. Do you run your model in a directory called "Model development"? The $ETAs generated would in theory do little harm, but it doesn't currently work with a space in the path.

I will mark this as a bug.

brunabernar commented 3 years ago

Worked with me, thanks! :)

This is because of two issues in PsN.

The $ETAs was added by PsN. This should only be done by default if you are using the linearize option, but it does it also for non-linear scm:s

I still think you have a space in your path. Do you run your model in a directory called "Model development"? The $ETAs generated would in theory do little harm, but it doesn't currently work with a space in the path.

I will mark this as a bug.

UUPharmacometrics / PsN

error in SCM #167