The appearance of this 2nd edition of the NM-TRAN Guide coincides with the appearance of NM-TRAN Version II Level 1. This version supports the new features of NONMEM Version IV Level 1 and PREDPP Version III Level 1. In addition, there are improvements in the level of support NM-TRAN provides to usage of NONMEM and PREDPP that does not take advantage of any new features of these latest versions.
The following is a list of changes of which an old NM-TRAN user may want to be particularly aware. Following each item in the list, a section in this Guide is cited (e.g. NM-TRAN V.C.4) where the item is discussed. The list is divided into two sections: items that apply in general and items that apply only when using PREDPP. With a item of the latter type, and when it may be useful to a reader of this Guide, a section in NONMEM Users Guide VI, the PREDPP Guide, is also cited (e.g. PREDPP III.F.6).
1. |
Generated code supports the use of conditional estimation methods. Such methods can be specified using options in the $ESTIMATION record. |
NM-TRAN III.B.14 |
2. |
Variables defined in the $PRED block become PRED-defined variables. Their values can be displayed in tables and scatterplots. There are ways of exempting variables from being PRED-defined. |
NM-TRAN III.B.7,B.16-17, IV.A |
3. |
Abbreviated code can be used when the Simulation Step is implemented. |
4. |
Verbatim code can be inserted into abbreviated code. Verbatim code is (any) FORTRAN code that is specially marked. It is taken by NM-TRAN and inserted verbatim into generated code, i.e. is inserted without there also being generated any additional code. This allows the user to use FORTRAN constructions not allowed in abbreviated code per se, e.g. CALL ROUT. |
NM-TRAN IV.I |
5. |
Restrictions on expressions involving ETA’s, EPS’s, or ERR’s are far fewer with Version II abbreviated code than are they with Version I abbreviated code. In particular, the exponential error model can be expressed. |
6. |
Abbreviated code recognizes the variable ICALL when it occurs in a logical expression being tested with an IF statement. This variable signals whether the subroutine is being called during simulation, or during data analysis, or (in the case of PRED) during an initialization or finalization phase. |
NM-TRAN IV.D |
7. |
The quantities appearing in NONMEM read-only commons ROCM11 and ROCM12 can be referenced as right-hand-quantities in abbreviated code. ROCM11 contains the quantities needed by the user to implement a mixture model. |
NM-TRAN IV.E.1 |
8. |
PRED error-recovery is supported with abbreviated code and with options on NM-TRAN control records. |
NM-TRAN IV.G, III.B.9,B.14 |
9. |
A new option on the $SIMULATION record, ONLYSIMULATION, allows a simulation to be implemented without any statistics based on the objective function also being computed. PRED-defined true-value variables are displayed using the simulated values for the @eta@ variables. Thus NONMEM is more easily used as a purely simulation tool. |
NM-TRAN III.B.13 |
10. |
The values of ETA’s can be displayed. These values are conditional estimates when a conditional estimation method is used. A $TABLE or $SCATTERPLOT record can contain the label ETAn (or ETA(n)), which labels values of ETA(n). |
NM-TRAN III.B.16-17 |
11. |
Two new options on a $TABLE record, NOHEADER and ONEHEARER, specify that either no headers occur in tables that are output to a Tables File, or that only one header occurs. |
NM-TRAN III.B.16 |
12. |
A new option on a $SCATTERPLOT record, ORD0, specifies that a line through 0 on the ordinate axis occurs in each of the specified scatterplots. |
NM-TRAN III.B.17 |
13. |
A new option on the $DATA record, CHECKOUT or CHECKDATA, can be used to implement the data checkout mode. With this mode tables and scatterplots can be requested, but nothing else, because abbreviated code is never executed. |
NM-TRAN III.B.5 |
14. |
A filename in the $DATA record can now be enclosed in quotes, allowing special characters to be used in the name. |
NM-TRAN III.B.5 |
15. |
Comment records can be included in an NM-TRAN data set. |
NM-TRAN II.C.3 |
16. |
With Version I the Data Preprocessor replaces a null data item in a NM-TRAN data set with a field of blanks in the NONMEM data set. With Version II this data item can be replaced with one consisting of a specified alphanumeric character. Thus if periods are preferred to blanks in the NONMEM data set, periods can replace blanks. |
NM-TRAN III.B.5 |
17. |
By default the Data Preprocessor attempts to produce a NONMEM data set comprised of 80 character (FORTRAN) records. It may need to suppress space between fields or generate multi-line NONMEM data records to accomplish this. A new option on the $DATA record, WIDE, specifies that only single-line NONMEM data records are to be generated. These records always include at least one blank between fields. They comprise a NONMEM data set which may be more legible and which can be processed by programs other than NONMEM. |
NM-TRAN III.B.5 |
18. |
The Data Preprocessor no longer allows completely blank lines in a NM-TRAN data set. Therefore, with a given data set error messages may arise with Version II where no such messages arise with Version I. |
19. |
Variables defined in the $PK block become PK-defined variables. Such variables can be used in the $ERROR block, and their values can be displayed in tables and scatterplots. There are ways of exempting variables from being PK-defined. |
NM-TRAN III.B.7, V.C.5-6 |
20. |
Variables defined in the $ERROR block become ERROR-defined variables. Their values can be displayed in tables and scatterplots. There are ways of exempting variables from being ERROR-defined. |
NM-TRAN III.B.7, V.C.6 |
21. |
PK-defined variables, in addition to the array elements P(n), can be used in the $DES and $AES blocks. When they do, they are called implicit basic parameters |
NM-TRAN V.C.7-9 |
22. |
With Version I the definition of a variable in an abbreviated code applies only to that code; it is a local definition. With Version II the definition applies to all abbreviated codes; it a global definition. This can lead to error messages with Version II if, say, a random variable is defined in one abbreviated code, and then redefined in a subsequent abbreviated code. Or, certain control streams which were appropriate with Version I may not produce the correct output with Version II. Or, compiler errors result when the name of a defined variable is that of an array whose elements symbolize left- or right-hand quantitites in another abreviated code. The use of COMRES=-1, as a pseudo-statement or as an option with the $ABBREVIATED record, restores the Version I behaviour. |
NM-TRAN III.B.7, IV.H |
23. |
A variable NEWIND is defined for PK and ERROR abbreviated codes analogously to the way this variable is defined for a PRED abbreviated code. |
PREDPP III.I NM-TRAN V.C.5-6 |
24. |
ALAGn is the recognized name of a new additional PK parameter, the absorption lag time for compartment n. |
PREDPP III.F.6 NM-TRAN V.C.5 |
25. |
PRED error-recovery from PK, ERROR, etc. is supported with abbreviated code and with options on NM-TRAN control records. |
PREDPP III.K, IV.F NM-TRAN III.B.9,B.14, IV.G, V.C.5-6 |
26. |
A new PK calling-protocol of PREDPP Version III, whereby PK is called with every event record, is now the default calling-protocol implemented with NM-TRAN version II. The calling-protocols of PREDPP Version II only allow more limited sequences of calls to PK (without also using CALL data items). As a result of this change, when there are event records in an individual record with the same time data item, but with different user-defined data items (not at all a common occurrence), results obtained using NM-TRAN Version II can differ from those obtained using Version I. The older calling-protocols are also implementable with NM-TRAN, using a pseudo-statement of the form CALLFL=n in the $PK block. With NM-TRAN Version II, in contrast to NM-TRAN Version I, there are no circumstances, other than when the pseudo-statement CALLFL=1 is used, in which NM-TRAN implements the particular calling-protocol that limits calls to PK to one per individual record. |
PREDPP III.H NM-TRAN V.C.5 |
27. |
With simple $ERROR blocks (e.g. $ERROR Y=F+ERR(1)), NM-TRAN still implements (by default) the calling-protocol that limits calls to ERROR to once per problem. With more complicated $ERROR blocks, a new ERROR calling-protocol of PREDPP Version III, whereby ERROR is called with every event record, is now the default calling-protocol implemented with NM-TRAN. The calling-protocols of PREDPP Version II only allow more limited sequences of calls to ERROR (without also using CALL data items). As a result of this change, when the abbreviated ERROR code modifies the prediction for an observation from the right-hand quantity symbolized by F (as with PD modeling), the predictions with nonobservation event records obtained using NM-TRAN Version II can differ from those obtained using Version I. The older calling-protocols are also implementable with NM-TRAN, using a pseudo-statement of the form CALLFL=n in the $ERROR block. With NM-TRAN Version II, in contrast to NM-TRAN Version I, there are no circumstances, other than when the pseudo-statement CALLFL=1 is used, in which NM-TRAN implements the particular calling-protocol that limits calls to ERROR to one per individual record. |
PREDPP IV.C NM-TRAN V.C.6 |
28. |
PK is called with every event record, or in other words, at every event time, but it can be called in addition at any time t an additional dose or lagged dose enters the system, by setting CALLFL in the $PK block to -2. With such a call, data items are available as right-hand quantities in the $PK block, as usual. Each DV, MDV, ID, L2, AMT, RATE, SS, II, ADDL, CMT, PCMT, CALL, EVID data item is that from the dose record. Each TIME data item is that from the first event record with time >t. Each other type of data item is by default that from the first event record with time >t. However, a new control record ($BIND) allows the user to choose, for each of these other types of data items, whether it is that from the dose record describing the dose, or the last event record with time <t, or the first event record with time >t. For further flexibility, every data item from the first of these records is always available at the call, but symbolized in a special way. The time t is also available as a right-hand-quantity. |
PREDPP III.B.2, III.H, III.I NM-TRAN V.C.5 |
29. |
A new record $TOL exists that allows TOL to be set differently between compartments. This may be useful with ADVAN9. |
PREDPP VI.D NM-TRAN V.C.10 |
30. |
Variables A(n), having the values of the amounts in the various compartments, are now defined for an ERROR abbreviated code. This allows easier modeling of pharmacodynamics and in general, decreases the need to use PCMT data items. |
PREDPP IV.D NM-TRAN V.C.6 |
31. |
When using ADVAN1-4 or ADVAN10, a variable SC is recognized that has the same meaning as Sn, where n is the number of the central compartment. This number, therefore, need not be remembered for the purpose of coding the scaling parameter for the central compartment. |
NM-TRAN V.C.5 |
Vertical lines appear in the right margin, marking changes or additions to the text that may be of particular interest to user of Version I.
In the first edition of this Guide the terms ’individual data’ and ’individual data set’ were used. In this edition these terms are replaced by ’single-subject data’ and ’single-subject data set’. The section clarifying the differences between population and single-subject data and between the use of @eta@ and @epsilon@ variables with these two kinds of data, section II.C.4, has been improved, and should be consulted.
TOP
TABLE OF CONTENTS
NEXT CHAPTER ...