[BUG] Using layout_umap and getting key error

Expected Behavior

a graph layout using UMAP for dim reduction to 2d

Actual Behavior

coming from layouts\auto.py KeyError: ‘2’

Example Code

Please see How to create a Minimal, Reproducible example for some guidance on creating the best possible example of the problem

import networkx as nx
from graspologic.layouts import layout_umap

g = nx.erdos_renyi_graph(10, 0.7, directed=False)
layout_umap(g)

Full Traceback

Paste the full traceback in case there is an exception

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
<ipython-input-17-1c062a6ca19a> in <module>
      3 
      4 g = nx.erdos_renyi_graph(10, 0.7, directed=False)
----> 5 layout_umap(g)

~\miniconda3\lib\site-packages\graspologic\layouts\auto.py in layout_umap(graph, min_dist, n_neighbors, max_edges, random_seed)
    177         min_dist=min_dist, n_neighbors=n_neighbors, random_state=random_seed
    178     ).fit_transform(tensors)
--> 179     positions = _node_positions_from(lcc_graph, labels, points, random_seed=random_seed)
    180     return lcc_graph, positions
    181 

~\miniconda3\lib\site-packages\graspologic\layouts\auto.py in _node_positions_from(graph, labels, down_projection_2d, random_seed)
    241     scaled_points = _scale_points(down_projection_2d, covered_area)
    242     partitions = leiden(graph, random_seed=random_seed)
--> 243     positions = [
    244         NodePosition(
    245             node_id=key,

~\miniconda3\lib\site-packages\graspologic\layouts\auto.py in <listcomp>(.0)
    246             x=scaled_points[index][0],
    247             y=scaled_points[index][1],
--> 248             size=sizes[key],
    249             community=partitions[key],
    250         )

KeyError: '2'

Your Environment

• Python version: 3.8
• graspologic version:

[bdpedigo]

totally fine with removing it from utils, personally!

[daxpryce]

@bdpedigo blech - we generate the edge list that leiden requires in utils.py where we generate a List[Tuple[str, str, float]]. That’s reasonable as a utility, but I think a Tuple[Dict[str, Any], List[Tuple[str, str, float]]] is not really something the common person is going to want - but it is what I need to do to fix this.

So the question is whether I can just axe this from utils or not. There’s benefits to having a “given a graph give me a basic edge list” but for all I know there’s like 15 different versions of that behavior already. So I figured I’d bring it up with you and see if you have a preference for what we do; axe it from utils or give everyone the Tuple[Dict[str, Any], List[Tuple[str, str, float]]] return type, which… doesn’t seem super useful. Or, I guess, replicate the function and have a private leiden specific version that does that.