Feature Requests
Closed this issue · 15 comments
Add a new InsertParameters class that works on the Copasi version 19 release. This is the only difference between the new and old version - that I know of.
Ensure other features of PyCoTools work with Copasi 19
Documentation should be updated. Make sure defaults are available for every kwarg. Include tables for kwargs that do the same thing.
Modify camel case to lower case where necessary. Be consistent with python conventions.
Change PEAnalysis to support seaborn, rather than just using bare matplotlib.
Create a PCA class in PEAnalysis
Need interface to the plotting features of PEAnalysis from multi-model and model selection stuff
Support for bifurcation analysis
Also update git markdown to include model selection features and include the changes you'll make to support copasi version 19.
Change write_item_template to write_config_template
Change PE.set_up to PE.setup()
Write a few extra lines in PyCoTools paper for model selection. Another figure?
Declare unicoding on each file
Fix tests to work with copasi 19
Create interactive plots using bokeh
Experiment with https://plot.ly/python/
Give each class str, repr and getitem methods
Replace GetModelQuantitities with Model
plot signal to noise ratio in peanalysisbto help visualize non identifiabilities
Create new parallel function for running copasi models in multi parameter estimation. Use same method as RunMultiParameterEstimation.setup1scan
Add features such as restricting parameter to certain experiments in PE setup file. Add the fact this is not supported in the documentation.
Include 1 optional parameter in PE.write_config_template() to override location of config file
Think about removing the use_config_start_values argument of parameter estimation
Update experiment mapper class to use dicts instead of string. Include affected experiments and options for duplicating parameters for each experiment
When using constraints, pycotools 1) does profile likelihood for the contraints as well and 2 changes the parameters.
This should be included in a testing suit for profile likelihoods