From: nonmem@optonline.net
Subject: [NMusers] omega and mixture
Date: Mon, 23 Oct 2006 10:37:50 -0400

 Hello NONMEM users,
 
 I try to combine a few different studies in one model:
 1.  IV infusion -  reach data (parameters: Cl, Vd);
 2.  IV infusion -  sparse data (parameters: Cl, Vd);
 3.  SC injection -  reach data (parameters: Ka, Cl, Vd);
 4.  SC injection -  sparse data (parameters: Ka, Cl, Vd);
 
 Between-subject variation in Ka is allowed for goup#3.  It is not 
 a problem to specify different differential equations and 
 different errors for each study.  Problem is that different 
 studies have different dimention of OMEGA matrix.  Thus, groups 1, 2, and 4 
 have $OMEGA BLOCK(2).  Group 3 has $OMEGA BLOCK(3).  I want to use 
 MIXTURE to specify 2 different $OMEGA matrices.  Is this a 
 reasonable and common approach?  

My code is 
$CONTR DATA=(STU)
$MIX 
  NSPOP=2
  IF (STU.EQ.1.OR.STU.EQ.6) THEN
    P(1)=1
    P(2)=0
  ELSE
    P(1)=0
    P(2)=1
  ENDIF 

where STU is a study number.  The code generates he following eror:

 AN ERROR WAS FOUND IN THE CONTROL STATEMENTS.

 AN ERROR WAS FOUND ON LINE  59 AT THE APPROXIMATE POSITION NOTED:
   IF (STU.EQ.1.OR.STU.EQ.6) THEN
       X
 THE CHARACTERS IN ERROR ARE: STU
 240  $MIX: DATA ITEM SHOULD BE LISTED WITH DATA OPTION ON $CONTR RECORD.
STOP 4 statement executed


Even though I specified the $CONTR record, I am getting the error.  
 
 Thank you!
 Pavel


_______________________________________________________