Merrick Strotton

Is an sha256 hash available for the current Qupath (0.3.2) release? (specifically the windows distribution?)

Thanks for the reply @petebankhead . Appreciate very much your efforts.

I have a large single cell set, where i'd love to speed up the Leiden clustering. Is it the igraph version of leiden implemented in scanpy? Just wondering if this...

So there are definite speed gains to be had with the igraph implementation of Leiden clustering. But the results are not exactly the same. I have run igraph straight from...

> The Leiden algorithm is now [included](https://igraph.org/python/doc/igraph.Graph-class.html#community_leiden) in the latest release of `python-igraph`, version 0.8.0. I believe this alleviates the need to depend on the `leidenalg` packages. The Leiden algorithm...

@vtraag @ivirshup I accidentally ran the `igraph_community_leiden` in a for loop, 4 times with the same settings and noticed the leiden output differed slightly each time. Is there a parameter...

For the 3k test dataset, introducing edge weights with ``` import random random.seed(1234) adjacency = sc._utils._choose_graph(adata, obsp=None, neighbors_key=None) g = sc._utils.get_igraph_from_adjacency(adjacency) clustering = g.community_leiden(objective_function='modularity', weights='weight') adata.obs['leiden_igraph_edge_weighted'] = pd.Series(clustering.membership, dtype='category', index=adata.obs.index)...

Setting `n_iterations=-1` in `g.community_leiden` certainly impacts run time (vs. default `n_iterations=2`), making runtimes more similar to `sc.tl.leiden()`. For large datasets though, run times with `g.comunity_leiden` still appear faster. The average...

Sorry, i made a critical typo in the time reports, where i listed the functions the wrong way round. I have updated the comment to correct this. To be clear....

Hi @vtraag @ivirshup, I have been unable to get the leiden clustering to run on a large 18.5 million cell dataset with either `sc.tl.leiden()` or the `python-igraph` implementation outlined above....