This PsN version has been tested on the following systems: Windows XP 32-bit with ActivePerl 5.16.3 and NONMEM 7.1.0, 7.1.2 and 7.2.0 Windows 7 64-bit with ActivePerl 5.16.3 and NONMEM 7.3.0 Windows 7 64-bit with Strawberry Perl 5.18.2 and NONMEM 7.2.0 and 7.3.0 MacOS X 10.6.6 with ActiveState Perl 5.16.3 and NONMEM 7.2.0 and 7.3.0 Ubuntu Linux 13.10 with Perl 5.14.2 and NONMEM 7.3.0 Ubuntu Linux 14.04 with Perl 5.18.2 and NONMEM 7.2.0 and 7.3.0 Fedora 20 with Perl 5.18.2 and NONMEM 7.2.0 Scientific Linux 6.3 with Perl 5.10.1 and NONMEM 7.2.0 and 7.3.0 CentOS 6.5 with Perl 5.10.1 and NONMEM 7.1.0 and 7.1.2 ************************ Installation ************************ Follow the installation instructions on http://psn.sourceforge.net/install.php ************************ Documentation ************************ All documentation is found in the doc subdirectory of the PsN installation directory. During installation it is also possible to have all documentation copied to a user-specified directory. All documentation can also be downloaded from the PsN website http://psn.sourceforge.net/ ************************ Dependencies ************************ - Xpose must, if used, be version 4.2.1 or later to handle PsN-3.5.3 or later output format. It is recommended to use the latest Xpose release from http://xpose.sf.net ************************ Known issues ************************ - Windows users need to be careful to choose a Perl version where the extra modules needed for PsN are available. See the instructions on http://psn.sourceforge.net/install.php - Installation problem on some Windows systems where perl 'system' command does not work. Make sure perl module File::Copy::Recursive is installed *before* (re)running the installation script setup.pl ********************************************* Changes and bug fixes from PsN 3.7.6 to 4.2.0 ********************************************* ___General___ - All modules have been converted from the old dia class system to Moose. All _subs.pm files have been converted into .pm files that can be used directly by Perl without the need for dia2code or fill_diacode.pl - All internal parameter checking is now done with MooseX::Params::Validate - PsN now comes bundled with a test suite used for verification. See the Testing section of developers_guide.pdf for instructions. - PsN now supports Strawberry Perl on Windows and ActivePerl on Mac - New common option -degree to all scripts, coupled to tweak_inits. Before, degree was hard-coded to 0.1, i.e. a perturbation of 10%, but now the user can choose the perturbation. Must be a number between 0 and 1. - tweak_inits will now perturb initial estimates by 'degree' relative to the original initial estimates, regardless of how many retries have been done. Before the perturbation would be larger and larger for every retry. - For runs with clean=0 a file called 'modelname' is created in each NM_run directory. This file contains the original name of the modelfile and can e.g. be used to set job name in wrapper scripts. The modelname file is removed if -clean is larger than 0. - NMQual8 support. Complete redesign, removed nmqual_xml option, handle .log.xml as standard output file from NONMEM that can be requested with -nm_output option. - PsN.pm is now treated as any PsN module and resides in the lib directory after installation - Redesign of tbs functionality, see common_options_defaults_version_psn.pdf. New options -tbs_lambda, -tbs_delta -tbs_zeta and -dtbs related to the tbs functionality. Bugfix: divide by zero in ccontra file for -tbs option - Versioned executables (for example "execute-4.2.0") now works on Windows. The setup script will automatically fix this also for older PsN versions installed in the same bin folder. - Handle Windows line breaks in model files and rawres_input files even if running on linux - Options -drop_dropped, -unwrap_table_files, -extra_data_files, -nmqual_xml, -run_on_sge_nmfe -run_on_lsf_nmfe and -wrap_data have been removed - Changed format of nnn_results.csv, e.g. sse_results.csv, to use 7 significant digits and either scientific or decimal notation. Before it was always 5 decimals in decimal notation. - All NM_run subdirectories will be created simultaneously before input is copied or any nonmem run is started. - More careful pre_compile_cleanup, leave cleaning of psn.lst and OUTPUT to nmfe. - Print a warning if old Mac text file (with only 0x0D as newline) format is used. - New document developers_guide.pdf with information for developers and testers - Cleaning in tool.pm, remove unused ForkManager - Bugfix in reading of matrices (covariance matrix, inverse cov matrix) from lst-file, make sure matrices are read and stored from the lst-file if files .cov and .coi are missing. - Bugfix in reading matrices from lst-file, remove NM7-tag before converting to MatrixReal - Bugfix data renumber_ascending and resamle, turn off redundant writes to data file. - Bugfix in handling of THETA/OMEGA/SIGMA labels. Sometimes PsN would crash because labels in model were not unique during updating of initial estimates. Now PsN will append a _ (underscore) to the label if necessary to make it unique. - Bugfix: Solved data bug where lines that did not start with a digit or a dot would be filtered out before bootstrapping/case deletion/statistics computations. PsN will instead use NONMEM's IGNORE sign, if set, to filter data files. - Bugfix: do not read final estimates from .ext if lst-file shows that NONMEM aborted the run - Bugfix in modelfit in verbose after cleaning > 2 - Bugfix in seed reset in modelfit based on NM_run number, could handle very large numbers - Bugfix: Corrected logic and run number dependence in delay code to avoid new process overload, max_fork_delay section - Bugfix: Fix regular expression in TBLN tag search to pick up a number and nothing else - Bugfix: Sometimes PsN runs on Windows got stuck in an infinite loop - Bugfix: Some Linux distributions (e.g. Fedora) needed the Digest::MD5 module for PsN to run. - Bugfix: some specific data sets (e.g. starting with a minus sign) could not be handled properly resulting in loss of data. ___Cluster/grid interaction___ - All job submission code has been rewritten. No perl process will be submitted to any cluster. Only nmfe or nmqual executions will be submitted. Code for each cluster/grid type is in its own class. - Option -slurm_project is renamed to -slurm_account (maps to sbatch option -A). - New option -slurm_partition which maps to sbatch -p - More robust slurm squeue commands to handle squeue error messages (e.g. Socket timed out) PsN will ask for completed, cancelled etc jobs using squeue options instead of checking for running jobs. This requires that the slurm option MinJobAge is not too small. - changes in slurm submit, submit entire bash script that first loops until psn.lst not visible (if retries are used), psn.mod is visible, and only then submit nmfe call. If invalid user id then try again a few times. - All output redirections will now be to nmfe_output.txt - Version of used nmfe script is written to version_and_option_info.txt ___NONMEM interaction___ - Support for new NONMEM 7.3 records $THETAI, $THETAR, $THETAP, $THETAPV, $OMEGAP, $OMEGAPD, $SIGMAP, $SIGMAPD - Order in which PsN prints records changed. $ABBREVIATED now comes after $DATA - Handle **** instead of subproblem number 10000 etc in lst-file when $SIM NSUBS>9999 in NONMEM 7.3 - Different default values for different NONMEM versions can be set in configuration file psn.conf. See psn_configuration.pdf - PsN will look for nonmem_mpi.exe and similar when checking compiler success when diagnosing missing lst-file - Changed syntax of option -nmfe_options to verbatim string instead of a comma separated list. See common_options_defaults_versions.pdf ___sse___ - New option -random_estimation_inits, see sse_userguide.pdf - New option -keep_tables, default is to remove tables. - NSUBS from pre-existing $SIMULATION will be deleted. - Better handling of failed simulations. If individual simulations fail then all estimations based on the corresponding simulated data set will be marked as failed, but sse will continue and compute results based on the remaining estimations. - Changed definition of skewness and kurtosis to standard definitions, see updated sse_userguide.pdf - Turn off redundant write to original data file copy in m1 - Bugfix: make sure common model_options are propagated to models when restarting an sse - Bugfix: in rawres input: allow third column to be model to handle new headers in sse. - Bugfix: in sse when simulating with prior, set the name of the file in which to write actual parameter values used for simulations ___scm___ - scm config file is copied to scm run directory (also for boot_scm and xv_scm) - Bugfix: handling more than 50 input columns in filter_data.mod, set SIZES PD - Bugfix: boot_scm will only use $INPUT for stratify_on, will ignore headers in datafile ___npc/vpc___ - Default name of rundirectory for a vpc is now vpc_dir - Around 40% faster autobinning - Parameters for binning algorithm are now easily changeable from within the code - Bugfix: Relax parameter checking when using bin_array on values for vpc - Bugfix: Make sure new model objects (e.g. after flip_comments) get copy of common_options::model_options - Bugfix: npc printing name of simulation model in output ___bootstrap___ - New option -keep_tables, default is to remove tables. ___mcmp___ - More clear error message when incorrectly defined model is incompatible with TABLE FIRSTONLY - Copying of old output to protect it when rerunning in existing run directory. ___cdd___ - Bugfix: redundant data writes have been turned off ___parallel_retries___ - New options -rawres_input, -offset_rawres, -in_filter and -samples to be used as alternatives to -min_retries. Function of options as in sse, except that inits are for estimation instead. See execute_userguide.pdf which also contains parallel_retries documentation. ___frem___ - Support for -start_eta option. See frem_userguide.pdf ___pvar___ - First release of new tool "parametric variability". See pvar_userguide.pdf ___linearize___ - Default name of rundirectory is now linearize_dir - Name of logfile changed to linlog.txt - Comment SCM-LINEARIZE-CONSTANTS have been removed from base model in linearize program ___nca___ - First release of new tool "noncompartmental analysis". See nca_userguide.pdf ___sir___ - First release of new tool "sampling importance resampling". See sir_userguide.pdf ___rawresults___ - First release of new tool "rawresults" that will create a raw results file from pairs of .mod and .lst files in the current directory. No userguide. ___update_inits___ - New userguide update_inits_userguide.pdf - can handle any file extension of the modelfile, not only .mod - can auto-renumber when file name is something like run123.mod - can now handle new NONMEM records (e.g. $THETAPD)