output_graph_picture output is not readable

[pikapika505]

Hello,

I saw this issue was discussed partially here output_visualization_CellNEST.py but the solution was for a previous release.

1. output_graph_picture on V1_Human_Lymph_Node_spatial data produces edge_graph.svg and edge_graph.pdf that look like this

2. Suggested solution of changing np.float -> np. float64 has already been done in the newer version of output_visualization_CellNEST.py. Changing it back to np.float did not change the output.

3. Additionally, when I run it on my own Visium data with ~150 000 cells it runs forever until interrupted by ctrlC then produces the same overlapped svg and no pdf.

This is one of the main functions that produces interpretable figures to explore the data. Could you please help me get proper results?

[pikapika505]

I think I figured it out. So, the actual problem was that write_dot() from NetworkX/PyGraphviz was overwriting or compressing the pos attribute during DOT export, resulting in tiny coordinates like pos="8,3!" in .dot file. I replaced this

write_dot(g, output_name + "_test_interactive_top"+ str(args.top_edge_count) +".dot")

in output_visualization_CellNEST.py with

import pygraphviz as pgv

A = nx.nx_agraph.to_agraph(g)

# Force correct spatial positions
for n in A.nodes():
    nid = int(n.get_name())
    n.attr["pos"] = f"{float(x_index[nid])},{float(-y_index[nid])}!"

    # Clear conflicting attributes safely
    for k in ("x", "y", "size", "physics"):
        n.attr[k] = ""

A.write(output_name + "_test_interactive_top" + str(args.top_edge_count) + ".dot")

And then re-ran cellnest visualize ... and cellnest output_graph_picture ... Now edge_graph.pdf for 4000 top edges looks like this