How to Extract Node Colors and node angles from NSBM (nsbm web viz tool)

graph-tool
1

I have python code to extract node positions, beizier points, and blocks after running the NSBM and export them. The exported data is later loaded by a custom JavaScript code that creates an interactive plot.

However, I’m facing a couple of challenges:

Extracting Node Colors: I really like the colors that Graph-tool generates in the plot, but I haven’t found a way to extract them for reuse. Currently, I’m generating random colors, but I’d prefer to keep the ones Graph-tool provides. Is there a workaround to access and export these colors?

Node Position Rotation/Reflection: The node positions seem to be rotated or reflected compared to the original Graph-tool plot. Is there a specific formula I can use to map these positions correctly and ensure they match the layout from Graph-tool?

graph-tool

text3
text31200×1198 352 KB

my result

image
image918×852 86.1 KB

my code

state_3 = minimize_nested_blockmodel_dl(g3gt, state_args=dict(overlap=False))
spos0, tg, tpos =state_3.draw(output="text3.png", beta=0.9)
cts = gt.draw.get_hierarchy_control_points(g3gt, tg, tpos, beta=0.9)
pos = g3gt.own_property(tpos)
pos = np.array(list(pos))

postg = np.array(list(tg.own_property(tpos)))
edgePos = list(cts)
dataEdge = {'beizer':[], 's':[], 't':[],}
nodes_graph = []
for b, e in zip(edgePos, g3gt.edges()):
    s = g3gt.vertex_index[e.source()]
    t = g3gt.vertex_index[e.target()]
    nodes_graph+=[s,t]
    dataEdge['beizer'].append(list(b))
    dataEdge['s'].append(s)
    dataEdge['t'].append(t)

dfEdge = pd.DataFrame(data=dataEdge)


numNodes = pos.shape[0]

numNodesBlocks = postg.shape[0] - numNodes

# memberships
st = [list(list(b)) for b in state_3.get_bs()]


sizes = [len(l) for l in st[1:len(st)]]

sizes0 = [0]+[val + np.sum(sizes[i+1:len(sizes)]) for i, val in enumerate(sizes)][::-1]


nodeIds = [
    list(range(int(sizes0[i])+1, int(s)+1))[::-1]
    for i, s in enumerate(sizes0[1:len(sizes0)])
][::-1]
nodeHierarchy0 = {}
for k, nodeLevel in enumerate(st[0]):
    nodeHierarchy0[k] = [nodeIds[0][nodeLevel]]
    for i in range(1, len(st)-1):
        newLevel = st[i][nodeLevel]
        nodeHierarchy0[k].append(nodeIds[i][newLevel])
        nodeLevel = newLevel

blocks = [nodeHierarchy0[k] for k in range(len(pos))]
1 Like