search

jonescompneurolab / hnn

The Human Neocortical Neurosolver (HNN) is a software tool that gives researchers/clinicians the ability to develop/test hypotheses on circuit mechanisms underlying EEG/MEG data.
http://hnn.brown.edu
Other
78 stars 22 forks source link

visual comparison of two different parameter files to easily see which parameters changed #221

Open stephanie-r-jones opened 4 years ago

stephanie-r-jones commented 4 years ago

In developing and testing hypotheses on the circuit origin of the data, it would be very useful for the user to have an easy way to see which parameters are different across different simulations, perhaps as sliders similar to the optimization dialog window.

jasmainak commented 4 years ago

what about parameters which are added or deleted? Do you want them to be part of the visualization?

stephanie-r-jones commented 4 years ago

Do you mean parameters that are added or deleted as the code expands? The parameters of interest would be those describing the local network connectivity, and the exogenous drive. For ERP simulations, then optimization tool optimizes over several parameters and these in particular would be important to visualize before and after optimization to see what changed.

jasmainak commented 4 years ago

no actually I was asking a more basic question. Since two parameter files don't always have the same set of parameters, I was asking whether the visualization would show only the common paramters or a set of predecided parameters that were also available in the files

rythorpe commented 4 years ago

This might help clarify the issue that @jasmainak is bringing up. Plotted below are the evoked input parameters pertaining to two different simulations. Note that the param file corresponding to the bottom image doesn't include values associated with the distal inputs that are in the top image.

laser_2020_02_12_3prox_4dist_params

laser_2020_04_08_3prox_params

rythorpe commented 4 years ago

The function that generated the plots above received an argument where the user specified which evoked inputs it would look for in both param files. This way, both figures would be side-by-side comparable. This should be relatively simple to implement in hnn_core.viz once we have refactored the hnn_core.feed module.

stephanie-r-jones commented 4 years ago

Ahh... I understand. I like these plots. However, rather than have the user have to input anything more specific other than the file names to compare, could the function just plot ALL parameters in EITHER file, and if a parameter is only in one file mark it as NAN in the corresponding space in the other file's plot?

jasmainak commented 4 years ago

One issue with plotting all the parameters is that the units and scale of different parameters may be vastly different.

rythorpe commented 4 years ago

One issue with plotting all the parameters is that the units and scale of different parameters may be vastly different.

This is true. In my situation I only plotted the parameters related to the evoked inputs, however, there are many other types of parameters and we will need to be creative with how we present them.

jasmainak commented 4 years ago

also the input times presented this way are not very meaningful in my opinion, it's better presented on the plot of the dipole. One idea could be to show the parameters such as gbar as connection strengths on a 3D "connectivity plot".

jasmainak commented 4 years ago

or a circle graph: https://mne.tools/stable/_images/sphx_glr_plot_mixed_source_space_connectivity_001.png