Fix issue 11712 "Function _plot_connectivity_circle connects nodes with score zero"

[aweinstein]

Reference issue

Fixes #11712 .

What does this implement/fix?

Now function _plot_connectivity_circle only shows links between nodes when the score between nodes is larger than zero.

[drammock]

just a note that I'm about to head on holiday for a week and wasn't able to review this beforehand. My apologies; hopefully another dev has time to look at it sooner. Otherwise I'll take a look sometime early in December.

[larsoner]

My only reservation with this is that I'm not sure we can guarantee a connectivity score of zero is meaningless, even if it is for most/many measures. For example if you had a measure that gave you the phase shift (in rad or deg) or group delay (in ms or whatever) between regions, zero is a reasonable value and you'd want to know the connection was instantaneous.

[aweinstein]

My only reservation with this is that I'm not sure we can guarantee a connectivity score of zero is meaningless.

You have the same issue if you use this function with the default parameter values: The black lines, corresponding to zero values, cannot be seen because they have the same color as the background.

We could add a threshold parameter (default value to zero) such that lines are plotted only if values are larger than this threshold.

[drammock]

You have the same issue if you use this function with the default parameter values: The black lines, corresponding to zero values, cannot be seen because they have the same color as the background.

Changing the colormap can address this. That said, I like the idea of having a threshold parameter. It's going to interact with the n_lines parameter, so we'd need to decide how to handle that (probably threshold should trump n_lines when they disagree, i.e., if n_lines is 20 but there are only 18 suprathreshold, we show 18. Also need to decide how to handle threshold when the connectivity measure can be positive/negative.