%load_ext autoreload
%autoreload 2
import matplotlib.pyplot as plt
from matplotlib.patches import Rectangle
import numpy as np
from flyvision import EnsembleView
from flyvision.utils import color_utils
ensemble = EnsembleView("flow/0000")
Loading ensemble: 0%| | 0/50 [00:00<?, ?it/s]
[2024-10-14 23:35:52] ensemble:166 Loaded 50 networks.
b
with ensemble.ratio(best=0.2):
fig, ax = ensemble.flash_response_index()
ymin, ymax = 0, 1
# to get locations of left most and right most T4 subtype ticks
xmin, xmax = [
p.get_position()[0] for p in ax.get_xticklabels() if p.get_text() in ["R1", "Tm3"]
]
# to place in between ticks
xmin -= 1 / 2
xmax += 1 / 2
xy = (xmin, ymin)
width = xmax - xmin
height = ymax
rect = Rectangle(xy, width, height, facecolor=color_utils.ON_FR, alpha=0.1)
ax.add_patch(rect)
ymin, ymax = 0, -1
# to get locations of left most and right most T4 subtype ticks
xmin, xmax = [
p.get_position()[0] for p in ax.get_xticklabels() if p.get_text() in ["L1", "Tm9"]
]
# to place in between ticks
xmin -= 1 / 2
xmax += 1 / 2
xy = (xmin, ymin)
width = xmax - xmin
height = ymax
rect = Rectangle(xy, width, height, facecolor=color_utils.OFF_FR, alpha=0.1)
ax.add_patch(rect)
ax.set_title("Flash response indices")
Text(0.5, 1.0, 'Flash response indices')
c
with ensemble.ratio(best=0.2):
fig, axes = ensemble.direction_selectivity_index()
ymin, ymax = 0, 1
# to get locations of left most and right most T4 subtype ticks
xmin, xmax = [
p.get_position()[0]
for p in axes[1].get_xticklabels()
if p.get_text() in ["T4a", "T4d"]
]
# to place in between ticks
xmin -= 1 / 2
xmax += 1 / 2
xy = (xmin, ymin)
width = xmax - xmin
height = ymax
rect = Rectangle(xy, width, height, facecolor=color_utils.ON, alpha=0.1)
axes[0].add_patch(rect)
# to get locations of left most and right most T4 subtype ticks
xmin, xmax = [
p.get_position()[0]
for p in axes[1].get_xticklabels()
if p.get_text() in ["T5a", "T5d"]
]
# to place in between ticks
xmin -= 1 / 2
xmax += 1 / 2
xy = (xmin, ymin)
width = xmax - xmin
height = ymax
rect = Rectangle(xy, width, height, facecolor=color_utils.OFF, alpha=0.1)
axes[1].add_patch(rect)
ax.set_title("Direction selectivity indices")
Text(0.5, 1.0, 'Direction selectivity indices')
d
from flyvision.analysis.flash_responses import (
flash_response_index,
fri_correlation_to_known,
)
from flyvision.analysis.moving_bar_responses import (
direction_selectivity_index,
dsi_correlation_to_known,
correlation_to_known_tuning_curves,
preferred_direction,
angular_distance_to_known,
)
with ensemble.ratio(best=0.2):
print(ensemble.names)
fris = flash_response_index(ensemble.flash_responses(), radius=6)
fri_corr = fri_correlation_to_known(fris)
['flow/0000/000', 'flow/0000/001', 'flow/0000/002', 'flow/0000/003', 'flow/0000/004', 'flow/0000/005', 'flow/0000/006', 'flow/0000/007', 'flow/0000/008', 'flow/0000/009']
with ensemble.ratio(best=0.2):
stims_and_resps_moving_edges = ensemble.moving_edge_responses()
# TODO: fix this, does not come out as expected
dsi_corr = dsi_correlation_to_known(
direction_selectivity_index(stims_and_resps_moving_edges)
)
tuning_corrs = correlation_to_known_tuning_curves(stims_and_resps_moving_edges)
t4_corrs = (
tuning_corrs.custom.where(cell_type=["T4a", "T4b", "T4c", "T4d"], intensity=1)
.median("neuron")
.squeeze()
)
t5_corrs = (
tuning_corrs.custom.where(cell_type=["T5a", "T5b", "T5c", "T5d"], intensity=0)
.median("neuron")
.squeeze()
)
pds = preferred_direction(stims_and_resps_moving_edges)
pd_distances = angular_distance_to_known(pds)
from flyvision.analysis.visualization.plots import violin_groups
fig, ax, *_ = violin_groups(
np.stack(
[
fri_corr.squeeze(),
t4_corrs.values,
t5_corrs.values,
dsi_corr.values,
],
axis=0,
)[:, None, :],
["FRI", "T4 tuning", "T5 tuning", "DSI"],
ylabel="correlation",
figsize=(1.8, 1.5),
ylim=(-1, 1),
colors=[
plt.get_cmap("Dark2")(0.125),
plt.get_cmap("Dark2")(0),
plt.get_cmap("Dark2")(0.25),
plt.get_cmap("Dark2")(0.375),
],
color_by="experiments",
scatter_edge_color="gray",
scatter_radius=5,
violin_alpha=0.8,
)
fig, ax, *_ = violin_groups(
pd_distances.values.flatten()[None, None, :],
["PD distance"],
ylabel="angular distance",
figsize=(1.8, 1.5),
ylim=(-1, 1),
colors=[
plt.get_cmap("Dark2")(0.5),
],
color_by="experiments",
scatter_edge_color="gray",
scatter_radius=5,
violin_alpha=0.8,
)
ax.set_ylim(
np.pi + 0.1,
-0.1,
)
(3.241592653589793, -0.1)
