Plotting in 3D¶
import numpy as np
import networkx as nx
import matplotlib
import matplotlib.pyplot as plt
%matplotlib notebook
import seaborn as sns
sns.set(font_scale=1.75)
sns.set_style("white")
# Palettes for edges and nodes
node_palette = sns.color_palette("dark")
edge_palette = sns.color_palette("pastel")
import random
np.random.seed(10)
from grapgh_io import GraphIO
ciona, ciona_e_att, ciona_n_att, ciona_g_att = GraphIO.load('../json_connectomes/ciona.json')
fafb, fafb_e_att, fafb_n_att, fafb_g_att = GraphIO.load('../json_connectomes/fafb.json')
worm, worm_e_att, worm_n_att, worm_g_att = GraphIO.load('../json_connectomes/worm_wiring_connectome_0_SL.json')
L = list(worm.edges)
L0 = []
L1 = []
L2 = []
for i in L:
if i[-1] == 0:
L0.append(i)
elif i[-1] == 1:
L1.append(i)
elif i[-1] == 2:
L2.append(i)
worm_0 = worm.edge_subgraph(L0)
worm_1 = worm.edge_subgraph(L1)
worm_2 = worm.edge_subgraph(L2)
def find_nodes(graph, name):
'''
Finds nodes in graph object that have coordinate or hemisphere annotations. Stores the coordinates,
or generates random coordinates that maintain L/R hemisphere dilineation.
'''
nodes_w_coors = []
nodes_w_sides = []
positions_axial = {}
positions_coronal = {}
y_pos = 1
x_left_pos = -20
x_right_pos = 35
for key in graph.nodes:
try:
if name == 'ciona':
if graph.nodes[key]['Side'] != '-':
nodes_w_sides.append(key)
if graph.nodes[key]['X'] != '-':
nodes_w_coors.append(key)
positions_axial[key] = np.asarray([graph.nodes[key]['X'], graph.nodes[key]['Y']])
positions_coronal[key] = np.asarray([graph.nodes[key]['X'], graph.nodes[key]['Z']])
elif name == 'worm':
if graph.nodes[key]['attr_dict']['hemisphere'] == 'left' :
nodes_w_sides.append(key)
positions_axial[key] = np.asarray([x_left_pos + random.uniform(-15, 15), y_pos])
positions_coronal[key] = np.asarray([x_left_pos + random.uniform(-15, 15), y_pos])
elif graph.nodes[key]['attr_dict']['hemisphere'] == 'right' :
nodes_w_sides.append(key)
positions_axial[key] = np.asarray([x_right_pos + random.uniform(-15, 15), y_pos])
positions_coronal[key] = np.asarray([x_right_pos + random.uniform(-15, 15), y_pos])
else:
continue
elif name == 'fafb':
if graph.nodes[key]['Side'] != False:
nodes_w_sides.append(key)
if graph.nodes[key]['X'] != False:
nodes_w_coors.append(key)
positions_axial[key] = np.asarray([graph.nodes[key]['X'], graph.nodes[key]['Y']])
positions_coronal[key] = np.asarray([graph.nodes[key]['X'], graph.nodes[key]['Z']])
y_pos += 1
except:
y_pos += 1
continue
return nodes_w_coors, nodes_w_sides, positions_axial, positions_coronal
ciona_coor_nodes, ciona_side_nodes, ciona_ax_coors, ciona_co_coors = find_nodes(ciona, 'ciona')
_, worm_side_nodes, _, _ = find_nodes(worm_0, 'worm')
_, worm_side_nodes, _, _ = find_nodes(worm_1, 'worm')
_, worm_side_nodes, _, _ = find_nodes(worm_2, 'worm')
fafb_coor_nodes, fafb_side_nodes, fafb_ax_coors, fafb_co_coors = find_nodes(fafb, 'fafb')
from mpl_toolkits.mplot3d import Axes3D
from matplotlib.lines import Line2D
def network_plot_3D(G, node_color, edge_color):
with plt.style.context(('ggplot')):
fig = plt.figure(figsize=(10,7))
ax = Axes3D(fig)
# Plot nodes
for key in G.nodes:
xi = G.nodes[key]['X']
yi = G.nodes[key]['Y']
zi = G.nodes[key]['Z']
ax.scatter(xi, yi, zi, c=node_color, edgecolors='k', alpha=0.7)
# Get starting and ending points for each edge
for i,j in enumerate(G.edges()):
x = np.array((G.nodes[j[0]]['X'], G.nodes[j[1]]['X']))
y = np.array((G.nodes[j[0]]['Y'], G.nodes[j[1]]['Y']))
z = np.array((G.nodes[j[0]]['Z'], G.nodes[j[1]]['Z']))
# Plot the edges as lines
ax.plot(x, y, z, c=edge_color, alpha=0.25)
# Legend
colors = [tuple(node_color[0]), edge_color]
texts = ["Soma", "Synapse"]
mkrs = ["o", "_"]
patches = [ plt.plot([],[], marker=mkrs[i], ms=10, ls="", mec=None, color=colors[i], label="{:s}".format(texts[i]) )[0] for i in range(len(texts)) ]
ax.legend(handles=patches)
def homophilic_plot_3D(G, node_color, edge_colors):
with plt.style.context(('ggplot')):
fig = plt.figure(figsize=(10,7))
ax = Axes3D(fig)
for key in G.nodes:
xi = G.nodes[key]['X']
yi = G.nodes[key]['Y']
zi = G.nodes[key]['Z']
ax.scatter(xi, yi, zi, c=node_color, edgecolors='k', alpha=0.7)
count = 0
for i,j in enumerate(G.edges()):
x = np.array((G.nodes[j[0]]['X'], G.nodes[j[1]]['X']))
y = np.array((G.nodes[j[0]]['Y'], G.nodes[j[1]]['Y']))
z = np.array((G.nodes[j[0]]['Z'], G.nodes[j[1]]['Z']))
# Plot edges, but each is color-coded specifically according to subtype
ax.plot(x, y, z, c=edge_colors[count], alpha=0.25)
count += 1
colors = [tuple(node_color[0]), edge_palette[1], edge_palette[0]]
texts = ["Soma", "Homophilic Synapse", "Heterophilic Synapse"]
mkrs = ["o", "_", "_"]
patches = [ plt.plot([],[], marker=mkrs[i], ms=10, ls="", mec=None, color=colors[i], label="{:s}".format(texts[i]) )[0] for i in range(len(texts)) ]
ax.legend(handles=patches)
edge_color = edge_palette[0]
node_color = np.asarray(node_palette[0]).reshape((1,3))
network_plot_3D(ciona.subgraph(ciona_coor_nodes), node_color, edge_color)
edge_color = edge_palette[1]
node_color = np.asarray(node_palette[1]).reshape((1,3))
network_plot_3D(fafb.subgraph(fafb_coor_nodes), node_color, edge_color)
