Pirana and PCluster: a modeling environment and cluster infrastructure for NONMEM

R.J. Keizer, M. van Benten, J.H. Beijnen, J.H.M. Schellens, AD Huitema

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Pharmacokinetic-pharmacodynamic modeling using non-linear mixed effects modeling (NONMEM) is a powerful yet challenging technique, as the software is generally accessed from the command line. A graphical user interface, Pirana, was developed that offers a complete modeling environment for NONMEM, enabling both novice and advanced users to increase efficiency of their workflow. Pirana provides features for the management and creation of model files, the overview of modeling results, creation of run reports and handling of datasets and output tables, and the running of custom R scripts on model output. Through the secure shell (SSH) protocol, Pirana can also be used to connect to Linux clusters (SGE, MOSIX) for distribution of workload. Modeling with NONMEM is computationally burdensome, which may be alleviated by distributing runs to computer clusters. A solution to this problem is offered here, called PCluster. This platform is easy to set up, runs in standard network environments, and can be extended with additional nodes if needed. The cluster supports the modeling toolkit Perl speaks NONMEM (PsN), and can include dedicated or non-dedicated PCs. A daemon script, written in Perl, was designed to run in the background on each node in the cluster, and to manage job distribution. The PCluster can be accessed from Pirana, and both software products have extensively been tested on a large academic network. The software is available under an open-source license.
Original languageEnglish
Pages (from-to)72-79
Number of pages8
JournalComputer Methods and Programs in Biomedicine
Volume101
Issue number1
DOIs
Publication statusPublished - 2011

Fingerprint

Dive into the research topics of 'Pirana and PCluster: a modeling environment and cluster infrastructure for NONMEM'. Together they form a unique fingerprint.

Cite this