def draw_network_interactive(dists_pd, GP, query_result, highlight, q):
G = nx.Graph()
for idx in range(len(dists_pd)):
G.add_edge(dists_pd.iloc[idx,0], dists_pd.iloc[idx,1], weight=dists_pd.iloc[idx,2],
quantile=dists_pd.iloc[idx,3])
width=min(900, int(np.sqrt(len(G.nodes)))*150)
if width < 400:
width=400
Gi = nx.Graph([(hit,hit) for hit in [highlight] if hit in G.nodes()])
highlight = hvnx.draw(Gi, GP, node_color="blue", node_size=3000, with_labels=False,
width=width, height=width)
Gi = nx.Graph([(hit,hit) for hit, rep in zip(query_result["Lead gene name"], query_result["Common lists"]) if rep==1 and hit in G.nodes()])
labels0 = hvnx.draw(Gi, GP, node_color="white", node_size=2500, with_labels=True,
width=width, height=width)
Gi = nx.Graph([(hit,hit) for hit, rep in zip(query_result["Lead gene name"], query_result["Common lists"]) if rep==2 and hit in G.nodes()])
labels1 = hvnx.draw(Gi, GP, node_color="lightgrey", node_size=2500, with_labels=True,
width=width, height=width)
Gi = nx.Graph([(hit,hit) for hit, rep in zip(query_result["Lead gene name"], query_result["Common lists"]) if rep==3 and hit in G.nodes()])
labels2 = hvnx.draw(Gi, GP, node_color="orange", node_size=2500, with_labels=True,
width=width, height=width)
if "Query" in query_result.columns:
Gi = nx.Graph([(hit,hit) for hit in np.unique(query_result["Query"]) if hit in G.nodes()])
labels3 = hvnx.draw(Gi, GP, node_color="red", node_size=2500, with_labels=True,
width=width, height=width)
else:
Gi = nx.Graph([(hit,hit) for hit in [query_result["Lead gene name"][0]] if hit in G.nodes()])
labels3 = hvnx.draw(Gi, GP, node_color="red", node_size=2500, with_labels=True,
width=width, height=width)
# .opts(tools=[HoverTool(tooltips=[('index', '@index_hover')])])
# edges
edge_styles = [{"edge_width": 8, "edge_color": "darkred", "style": "solid", "alpha": 1}, # 0.1%
{"edge_width": 8, "edge_color": "darkred", "style": "solid", "alpha": 1}, # 0.5%
{"edge_width": 8, "edge_color": "darkred", "style": "solid", "alpha": 1}, # 1%
{"edge_width": 6, "edge_color": "red", "style": "solid", "alpha": 1}, # 2.5%
{"edge_width": 6, "edge_color": "red", "style": "solid", "alpha": 1}, # 5%
{"edge_width": 4, "edge_color": "darkorange", "style": "solid", "alpha": 1}, # 10%
{"edge_width": 3, "edge_color": "#ababab", "style": "solid", "alpha": 1}, # 25%
{"edge_width": 2, "edge_color": "lightgrey", "style": "solid", "alpha": 1}, # 50%
{"edge_width": 2, "edge_color": "lightgrey", "style": "dotted", "alpha": 1}] # 100%
edges = []
for q_cat, style in zip(q.index[1:], edge_styles):
edgelist = [(u, v) for (u, v, d) in G.edges(data=True) if d['quantile']==float(q_cat)]
if len(edgelist) == 0:
continue
Gi = nx.Graph(edgelist)
edge = hvnx.draw(Gi, GP, node_size=2000, with_labels=False,
width=width, height=width, **style)
edges.append(edge)
nw = hv.Overlay(edges[::-1]) * highlight * labels0 * labels1 * labels2 * labels3
return nw
def draw(self, **kwargs: Dict) -> Overlay:
"""Draws the graph and returns the hvplot object.
Parameters
----------
**kwargs : Dict[str, Any]
keyword arguments given to the hvplot.networkx.draw method
Returns
-------
Overlay
HoloViews Overlay representing the correlation graph
"""
return hvnx.draw(self.graph,
pos=self.pos,
edge_width="weight",
node_color="cluster_corr",
labels="name",
colorbar=True,
**kwargs)
def draw_graph_hv(G):
pos = nx.fruchterman_reingold_layout(G)
edges = G.edges()
colors = [G[u][v]['color'] for u, v in edges]
weights = [G[u][v]['weight'] for u, v in edges]
options = {
'node_size': 2000,
'node_color': '#A0CBE2',
'edge_width': weights,
'edge_color': colors,
'width': 1800,
'height': 900,
'with_labels': True,
'alpha': 0.8
}
g = hvnx.draw(G, pos, **options)
renderer = hv.renderer('bokeh')
renderer.save(g, 'graph.html')
plot = renderer.get_plot(g).state
output_file("graph.html")
save(plot, 'graph.html')
show(plot)
def connections(neuron='BAGL', connstyle=CONN_STYLE):
'''Prepare graph for plotting'''
G = nx.DiGraph()
syn_in = dfs.index[dfs[neuron] > 0].tolist()
intens_in = dfs.loc[dfs[neuron] > 0, neuron]
syni = [(pre, neuron, {}) for pre in syn_in]
G.add_edges_from(syni)
gaps_ = dfg.index[dfg[neuron] > 0].tolist()
intens_g = 0.1 * dfg.loc[dfg[neuron] > 0, neuron]
gaps = [(neuron, k, {}) for k in gaps_] + [(k, neuron, {}) for k in gaps_]
G.add_edges_from(gaps)
syn_out = dfs.T.index[dfs.T[neuron] > 0].tolist()
intens_out = dfs.T.loc[dfs.T[neuron] > 0, neuron]
syno = [(neuron, post, {}) for post in syn_out]
G.add_edges_from(syno)
G.remove_node(neuron)
pos = nx.layout.circular_layout(G, scale=2)
G.add_node(neuron)
pos[neuron] = np.array([0, 0])
def draw_nodes(shape='o', category=inters, col='k'):
hvnx.draw_networkx_nodes(
G,
pos,
node_color=col,
nodelist=[n for n in G.nodes if n in category],
node_size=NODE_SIZE,
alpha=0.9)
draw_nodes(SENSOR_SHAPE, sensors, SENSOR_COLOR)
draw_nodes(INTER_SHAPE, inters, INTER_COLOR)
draw_nodes(MOTOR_SHAPE, motors, MOTOR_COLOR)
nx.draw_networkx_labels(G, pos, font_color='w', font_weight='bold')
nx.draw_networkx_edges(G,
pos,
arrowstyle=style,
edgelist=syni,
edge_color='g',
connectionstyle=connstyle,
arrowsize=10,
alpha=0.7,
width=intens_in,
node_size=NODE_SIZE)
nx.draw_networkx_edges(G,
pos,
arrowstyle=style,
edgelist=syno,
edge_color='r',
connectionstyle=connstyle,
arrowsize=10,
alpha=0.5,
width=intens_out,
node_size=NODE_SIZE)
nx.draw_networkx_edges(G,
pos,
arrowstyle=styleg,
edgelist=gaps,
edge_color='Gold',
arrowsize=10,
alpha=0.8,
width=np.hstack((intens_g, intens_g)),
node_size=NODE_SIZE)
#plt.axis('off')
hvnx.draw(G, pos)
hvnx.show()
from ndlib.viz.bokeh.DiffusionPrevalence import DiffusionPrevalence
import matplotlib as plot
import selenium
import hvplot.networkx as hvnx
from itertools import compress
import pandas as pd
import random
from ndlib.viz.mpl.TrendComparison import DiffusionTrendComparison
#%% PREPROCESSING:
# Load graphs and remove isolated nodes to only study the main
# connected component.
# LOAD GRAPH FOR MAIN NETWORK (SCENARIO 1)
raw_net = nx.read_graphml('budapest_large.graphml')
raw = hvnx.draw(raw_net, node_color = 'lightblue')
print('Properties raw_net: '+ str(len(raw_net.nodes()))
+ ' and ' + str(len(raw_net.edges())))
# Remove isolated nodes
main_net = raw_net
main_net.remove_nodes_from(list(nx.isolates(main_net)))
processed = hvnx.draw(main_net, node_color = 'lightblue')
print('Properties main_net: '+ str(len(main_net.nodes()))
+' and '+ str(len(main_net.edges())))
# LOAD NETWORK SCENARIO 2
deg_net = nx.read_graphml('budapest_large.graphml')
deg_net.remove_nodes_from(list(nx.isolates(deg_net)))
# LOAD NETWORK SCENARIO 3
node_size_list = list(node_sizes.values())
number_of_names_to_display = 10
add_labels = sorted(node_sizes, key=node_sizes.get,
reverse=True)[:number_of_names_to_display]
label_dict = {name: name if name in add_labels else "" for name in K.nodes}
# In[9]:
hvnx.draw(
I,
pos,
height=800,
width=800,
node_size=node_size_list,
arrowhead_length=0.01,
with_labels=True,
labels=label_dict,
edge_width='weight',
edge_alpha=0.7,
edge_color='brown',
node_color='eigenvector_centrality',
node_cmap=plt.cm.plasma,
)
# # Cliques
# Cliques are sets of nodes that are all connected to each other. That is, if we have four nodes, A, B, C, and D, all possible connections exist:
# > A --- B</br>
# > A --- C</br>
# > A --- D</br>
# > B --- C</br>
# > B --- D</br>
def test_nodes_are_not_sorted(self):
plot = hvnx.draw(self.g)
assert all(self.nodes == plot.nodes.dimension_values(2))
for period in enumerate(grphdict.keys()):
n = period[0]
pn = period[1]
periodgraph, pos = period_graph(pn)
labels = {node: node for node in periodgraph.nodes() if node in commonnames[pn]}
start = periods[pn]['start']
end = periods[pn]['end']
chart.title = f"REMP network {start}-{end}"
chart = hvnx.draw(G=periodgraph,
pos=pos,
with_labels=True,
labels=labels,
node_size=hv.dim('centrality')*200,
node_color='entity_type',
edge_color='community',
cmap='accent',
edge_cmap='category20',
node_tooltip=['name'],
#node_label='name',
font_color="black",
#font_size='11',
width=800,
height=600,
)
#chart.configure_view(width=800, height=600,)
chart.Overlay.opts(title=f"REMP network {start}-{end}")
if not charts:
charts = chart
else:
charts = charts + chart