Producing more colors in sensitivity analysis

[aryaman3900]

Hi @himoto ,

I am Aryaman Joshi, currently working over the EGFR-Ras signaling network using biomass.

I am trying to perform sensitivity analysis and am getting the following error:

len(sensitivity_options['cmap']) must be equal to or greater than len(problem.conditions)

I understand there is a requirement of producing colors equal to the conditions in my model (51) and tried to deal with it in a similar manner as done in SingleObservable() in viz.py. However, upon doing so I realized that sensitivity_options is not subscriptable as SensitivityOptions() accepts only one parameter.

Could you please let me know how to produce different colors for my model conditions?

Regards,
Aryaman

[himoto]

Hi @aryaman3900,

Thank you very much for using biomass!
If I remember correctly, you produced 51 colors for SingleObservable() via:

def get_single_observable_options(self):

    cm_name = 'jet'
    get_cmap = plt.cm.get_cmap(cm_name)
    colors = get_cmap(np.linspace(0, 1, 51))
    for i, _ in enumerate(self.obs_names):
        self.single_observable_options[i].cmap = [colors[j] for j in range(51)]
        self.single_observable_options[i].shape = list(Line2D.filled_markers) * 4
    return self.single_observable_options

To use the same colormap for sensitivity analysis, you need to modify get_sensitivity_opttions as follows:

def get_sensitivity_options(self):

    cm_name = 'jet'
    get_cmap = plt.cm.get_cmap(cm_name)
    colors = get_cmap(np.linspace(0, 1, 51))
    self.sensitivity_options.cmap = [colors[j] for j in range(51)]
    return self.sensitivity_options

I think this would work, but currently it is very difficult to visualize the result of sensitivity analysis as barplot when a model contains dozens of simulation conditions. I recommend you to start from around 5 conditions.

Please do let me know if the problem still persists.

Best,
Hiroaki

[aryaman3900]

@himoto Yes, this works. I was making an error in writing the code. Thank You very much.

Kind regards,
Aryaman