Customize a plot

The SpikeInterface widgets are designed to have reasonable default plotting options, but sometimes you’ll want to make adjustments to the plots. The plotting functions all return a Widget object. These contain and give you access to the underlying matplotlib figure and axis, which you can apply any matplotlib machinery to. Let’s see how to do this in an example, by first making some synthetic data and computing extensions which can be used for plotting.

import spikeinterface.full as si
import matplotlib.pyplot as plt

recording, sorting = si.generate_ground_truth_recording(seed=1205)
sorting_analyzer = si.create_sorting_analyzer(sorting=sorting, recording=recording)
sorting_analyzer.compute({"random_spikes": {'seed': 1205}, "templates": {}, "unit_locations": {}})

unit_locations = sorting_analyzer.get_extension("unit_locations").get_data()

estimate_sparsity (no parallelization):   0%|          | 0/10 [00:00<?, ?it/s]

estimate_templates_with_accumulator (no parallelization):   0%|          | 0/10 [00:00<?, ?it/s]

Now we can plot the unit_locations and unit_templates using the appropriate widgets (see the full list of widgets for more!). These functions output a Widget object. We’ll assign the unit locations widget to fig_units.

fig_units = si.plot_unit_locations(sorting_analyzer)

# Each widget contains a `matplotlib` figure and axis:
print(type(fig_units.figure))
print(type(fig_units.ax))

<class 'matplotlib.figure.Figure'>
<class 'matplotlib.axes._axes.Axes'>
../_images/customize_a_plot_4_1.png

By gaining access to the matplotlib objects, we are able to utilize the full matplotlib machinery: adding custom titles, axis labels, ticks, more plots etc. Let’s customize our unit locations plot. (Note: the SpikeInterface Team does not endorse the following style conventions):

# Get the widget
fig_units = si.plot_unit_locations(sorting_analyzer)

# Modify the widget's `axis`` to set the title and axes labels
fig_units.ax.set_title("My favorite units", fontname = "Comic Sans MS")
fig_units.ax.set_xlabel("x probe location (um)")
fig_units.ax.set_ylabel("y probe location (um)")

# You can also set custom ticks
fig_units.ax.set_xticks([-60,-30,unit_locations[0,0],30,60])
fig_units.ax.set_xticklabels([-60,-30,"unit_0_x",30,60])
fig_units.ax.set_yticks([-40,-20,0,unit_locations[0,1],40])
fig_units.ax.set_yticklabels([-40,-20,0,"unit_0_y",40])

# Change the limits of the plot
fig_units.ax.set_xlim((-30,50))
fig_units.ax.set_ylim((-50,50))

# And add extra information on the plot
fig_units.ax.text(unit_locations[6,0], unit_locations[6,1]+5, s="UNIT 6!!!", fontname="Courier")

fig_units

<spikeinterface.widgets.unit_locations.UnitLocationsWidget at 0x147a81520>
../_images/customize_a_plot_6_1.png

Beautiful!!!

You can also combine figures into a multi-figure plot. The easiest way to do this is to set up your figure and axes first, then tell SpikeInterface which axes it should attach the widget plot to. Here’s an example of making a unit summary plot.

import matplotlib.pyplot as plt
fig, axs = plt.subplots(ncols=2, nrows=1)

unit_id=8
si.plot_unit_locations(sorting_analyzer=sorting_analyzer, ax=axs[0])
si.plot_unit_templates(sorting_analyzer, axes=[axs[1]], unit_ids=[f'{unit_id}'])

axs[0].plot([unit_locations[8,0], unit_locations[8,0]+50], [unit_locations[8,1], unit_locations[8,1]+50])
axs[0].text(unit_locations[8,0]+52, unit_locations[8,1]+52, s=f"Unit {unit_id}")
axs[0].set_title("Unit location", fontsize=10)

fig.suptitle(f"Unit {unit_id} summary", fontfamily="Comic Sans MS", fontsize=20)

fig.tight_layout()
../_images/customize_a_plot_8_1.png

For more details on what you can do using matplotlib, check out their extensive documentation