Beyond the Kuiper Belt

• added support for Copasi Optimisation task. This also uses the -e option.
• bugfix: added is_package_installed.r to MANIFEST.ini.
• SBpipe v4.18.0, sbpiper v1.8.0, sbpipe_snake v1.0.0 and above are released under MIT License.
  Previous versions of these packages were released under GNU GPL v3.
• Improved project consistency and added warnings.
• Snakemake files moved to a separate repository (https://github.com/pdp10/sbpipe_snake.git).
• Added script for moving data sets and update indexes.
• Project and documentation clean-up.
• SBpipe is now available on pypi.org.
• Improved setup.py file for python packaging
• SBpipe tests no longer require CopasiSE.
• Documentation update.
• Added generate_tarball option to all the remaining pipelines in native SBpipe.
• Improved output messages.
• Added progress information for native SBpipe.
• Added PCA analysis for the best parameter estimates. Replaced conda channel "r" with "conda-forge".
• Improved data analysis scalability for parameter estimation (using Snakemake).
• Added checks whether a COPASI model can be loaded and executed correctly. This is based on Python bindings for COPASI.
• Optimisation of snakemake pipelines. Improved efficiency of rules for analyses.
• Bugfix - SGE and LSF job names now include a random string, avoiding potential interactions among multiple
  SBpipe executions. Whilst this does not affect the results, it was still a performance-related bug.
• SBpipe R code is now an independent R package called sbpiper. This is imported by SBpipe as
  an external dependency. Users can invoke SBpipe functions for data analysis directly from their R code.

Mars

This release includes the following features / updates:

• added option exp_dataset_alpha to sim pipeline. This option allows to plot experimental
  data with an alpha level.
• data analysis for sim pipeline is scalable.
• improved yaml files for installing SBpipe using conda. SBpipe is now tested on Python 2.7 and 3.6.
• added transparencies and improved simulation plots combined with data set.
• added release.sh script for releasing SBpipe versions automatically.
• if data_point_num is [0, est_param_number], the analysis task for parameter estimation will
  continue BUT the thresholds will be discarded.
• bug fix - conda build package after conda was upgraded to v3.x.x
• bug fix - constraints in parameter estimation using Copasi
• added scripts
• code optimisation for parameter estimation pipeline.
• Improved conda packaging.
• Improved import of parameter names for parameter estimation pipeline.
• Improved SBpipe packaging (snakemake is not a requirement)
• SBpipe pipelines are also available as snake files. Therefore, SBpipe can be run using Snakemake.
• SBpipe is now also available as conda package (installation+dependencies: conda install -c pdp10 sbpipe)
• The environment variable SBPIPE is no longer necessary.
• Anaconda can be used for installing SBpipe dependencies. This improves portability on Linux
  and Windows OS.
• subprocess.Popen() and logging fileConfig() use with .. as ..: construct with Python3+.
• changed chi^2 label to obj val in parameter estimation plots
• Added additional arguments to sbpipe
• Output is now coloured.
• Improved logging messages. Added log.debug() calls.
• Added sbpipe() function in main.py to facilitate programmatic use of sbpipe.
• Replaced Python getopt with argparse.
• Improved unit tests and nosetests with Travis-CI.
• Replaced INI configuration files with YAML configuration files.
• Skip heatmap and multiple time course plot if only one simulation is run. These plots are
  just redundant.
• removed support for running R, Octave, and Java models directly as these can be run via
  a Python model wrapper.
• bug fixes

Earth

This release includes the following features / updates:

• added prints to r plotting functions
• pipeline analyses are executed on cluster (local, sge, lsf) using sbpipe parcomp module.
• renamed option pp_cpus to local_cpus, after removal of parallel python.
• renamed value pp to local for option cluster after removal of parallel python.
• added support for Python 3. The code is now expected to work for Python 2.7+, 3.2, and 3.6.
• replaced parallel python with python multiprocessing package. This should facilitate the transition to Python 3.
• removed deprecated source code for manually randomising parameters before parameter estimation in Copasi files.
• Copasi and PL-based simulators now share a large amount of code.
• adapted programming language-based simulators to use ps1 and ps2 post-processing code. All simulators support all the pipelines.
• moved post-processing code for ps1 and ps2 from Copasi to Simul.
• improved code cohesion by moving utility code into Simul class().
• improved output name consistency for report and plot files
• improved sorting of plots in latex/pdf report for ps1 pipeline
• added test case for stochastic double parameter scan.
• double parameter scans can be executed in parallel as repeats.
• added support for stochastic double parameter scans.
• added test case for stochastic single parameter scan.
• single parameter scans can be executed in parallel as repeats.
• added support for stochastic single parameter scans.
• modularised parallel computation within Copasi simulator
• added heatmap plot representing stochastic repeats for the time course simulation.
• moved R code from pipelines to R/
• redesign of simulation plots. Improvements plus based on melt function.
• removed remaining old gplots dependent code. Sbpipe only uses ggplot2 now.
• added two new plots useful for stochastic simulation
• improved reuse for all R plots.
• added plots reproducing all the single simulations per species.
• changed main script name from run_sbpipe.py to sbpipe.py.
• Added support for parameter estimation using non-Copasi models. Test using R model.
• Skip Java, Python, and R model tests if their dependencies are not satisfied.
• Optimised Java, Python, and R simulators. Report file names are passed as input argument. Models do not need to be replicated.
• Java models can be used for model simulation in addition to Copasi.
• Python models can be used for model simulation in addition to Copasi.
• R models can be used for model simulation in addition to Copasi.

Venus

This release includes the following features / updates:

• improved threshold levels for Sampled PLE plots.
• added 20 tests including wrong configuration file settings.
• extensive refactory of unit tests.
• source code uses PEP8 standard
• source code cleaning and reformatting.
• improved source code by eliminating some warning highlighted by PyCharm.
• moved script core functions within sbpipe package.
• the copasi package is now a dynamically loaded simulator. Users can choose the simulator
  to use in the configuration file.
• simulators are loaded dynamically. Uncoupling between simulators and pipelines.
• separation of code for generating data from pipeline package.
• improved source code modularisation for the whole program.
• extracted scripts (run_sbpipe and cleanup_sbpipe) from sbpipe/.
• sbpipe supports execution as a Python module (main.py).
• all Python imports are now absolute (in agreement with Python 3).
• improvements to program prints
• project renamed sbpipe
• added AIC, AICc, BIC to the parameter estimation summary table.
• randomisation of initial parameter values for parameter estimation is now only
  performed by Copasi.
• added plots comparing model simulation vs experimental data in simulate pipeline.
• improving plot margins for simulate and single parameter scan pipelines.
• source code refractoring in parameter estimation analysis task.
• fixed a bug in parameter estimation pipeline related to the filtering of confidence
  intervals from the complete data set.
• added ratios in parameter estimation summaries to investigate the distance between the
  estimated parameter and its confidence intervals.
• plot polishing.
• separated options for plotting 2d correlations within 66%, 95%, or 99% confidence intervals.
• added 99% confidence intervals parameter estimation plots.
• added option to plot parameter estimation plots using the scientific notation.
• improved plots layout (fonts, legends).
• added option for y axis label to simulate and single parameter scan pipelines.
• Copasi models are fully consistent.
• table of estimated parameter and confidence values is in normal scale (not log10).

Mercury

This release includes the following features / updates:

• completed source code documentation
• completed user and developer manuals.
• configured Python Sphinx for documenting SB pipe
• bug fixes.
• separation of pdf report code from pipelines.
• configuration sessions integrated in pipeline classes.
• pipelines converted to classes.
• added option for plotting parameter estimation results in log10 parameter space (default).
• improved heat palette for double parameter scan and coloured scatterplots.
• added test files for double parameter scan
• ported all Matlab code to Python / R
• added pipeline for double parameter scan (parsing, plots, report)
• further removal of deprecated files
• generated copasi files for parameter estimation now moved to Working_Folder/xx/
• improved insulin receptor model for testing.
• Copasi report files now in Models/ .
• Copasi experimental data files now in Models/ .
• added scripts for automatically installing Python and R package dependencies.
• use of sections in configuration
• separation of configuration file parsing from program logic.
• restructuring dataset parsing for simulate and single_param_scan.
• added parameter scan plot with homogeneous lines (useful for plotting param conf. interv.).
• replaced all prints with Python logging.
• improved LaTeX reports
• tested parameter estimation using Gillespie algorithm for model simulation.
• configured Travis-CI for continous integration tests.
• pipeline renaming.
• added computation for parameter confidence intervals.
• added plot for fit history.
• added 2D parameter correlations using 66% or 95% confidence levels from calculated PLE.
• added profile likelihood estimation based on intermediate estimations.
• cleaned pipeline output.
• added documentation for configuring Copasi.
• removed part of the deprecated code.
• internalised code for each pipeline; run_sb_pipe.py is the main executor for sb_pipe.
• bug fixes.
• models can now be simulated in parallel using PP, SGE, or LSF.
• separation of parallel code from param_estim__copasi pipeline. It is generic now.
• sb_pipe should now be platform independent (untested yet).
• removed unused dependencies.
• better separation of test cases.
• pipeline steps can be executed separately.
• pipeline restructuring (separation of the steps: generate data, analyse data, and generate report).
• model parameters can now be estimated in parallel using PP, SGE, or LSF.
• removed old deprecated code.
• restructuring source code in the lib/ folder (now sb_pipe/pipelines and sb_pipe/utils).
• finalised skeleton for sb_param_estim pipeline.
• added parameter correlation plots for sb_param_estim pipeline.
• ported R gplots code to ggplot in sb_param_scan__single_perturb pipeline.
• ported R gplots code to ggplot in sb_simulate pipeline.
• sb_pipe is now a Python package.
• added documentation (readme, developer_guide).
• added unit tests and setup.py.
• ported Bash / sed / grep and cut code to Python in sb_param_estim pipeline.
• ported Bash / sed / grep and cut code to Python in sb_param_scan__single_perturb pipeline.
• ported Bash / sed / grep and cut code to Python in sb_simulate pipeline.
• added param_estim__copasi.sh.
• improved configuration file.
• simulation time start, end, xaxis label and time step now replace the parameter team.
• adjusted sb_simulate.sh, sb_param_scan__single_perturb.sh, sb_sensitivity.sh.
• packaging of sb_modules in /bin.
• added test scripts.