Dear NMusers,
For those of you using the Windows-based version of the WAM algorithm
(Wald's Approximation Method for covariate model building) developed by
Pharsight based on the methodology in the manuscript by Kowalski &
Hutmacher, JPP 2001;28:253-275, please note that we recently determined that
there is a bug with WAM v1.0. A new version, WAM v1.01, is now available
for download from the NONMEM repository
(ftp://ftp.globomaxnm.com/Public/nonmem/wamv1.01/) that corrects this bug.
The bug with WAM v1.0 occurs when there are one or more THETAs that are held
FIXED in the NONMEM control stream for the full model and the COV step
output is COMPRESSED as a result of the user specifying the COMPRESS option
on the $COV step or when NONMEM automatically invokes the COMPRESS option
when reporting COV step output for large matrices. In this setting, WAM
v1.0 will incorrectly construct the covariance sub-matrix for the covariate
parameter estimates that it uses in the matrix computations for Wald's
approximation to the likelihood ratio test. WAM users are advised to
download WAM v1.01 from the NONMEM repository which corrects this bug. My
apologies for any inconvenience this may cause.
Please note that Pharsight owns the copyright to this freeware but has
kindly allowed A2PG to provide a fix for this bug. There are no plans to
continue to maintain this software even though Pharsight has graciously
given A2PG permission to do so. We anticipate that there will be
compatibility issues with using WAM v1.01 with NONMEM VII and hence this
freeware version may have a limited shelf life. We encourage software
developers to consider implementing this methodology as another modeling
tool/utility in their model-building software platforms. A2PG would be
happy to work with any developers who may be interested.
Also note that this bug only applies to the Windows-based version of the WAM
algorithm which integrates the WAM computations with the running of NONMEM.
The Splus and SAS versions for performing the WAM computations independent
of the software used for estimation (e.g., NONMEM) do not suffer from this
bug. For those who may be interested in the Splus or SAS versions please
send your requests to me by email.
Kind regards,
Ken
Kenneth G. Kowalski
President & CEO
A2PG - Ann Arbor Pharmacometrics Group, Inc.
110 E. Miller Ave., Garden Suite
Ann Arbor, MI 48104
Work: 734-274-8255
Cell: 248-207-5082
Fax: 734-913-0230
[email protected]