infercnvpy.pl.chromosome_heatmap doesn't pass dendrogram data to scanpy.pl.heatmap

[Jura Pintar]

Report

When infercnvpy.pl.chromosome_heatmap and infercnvpy.pl.chromosome_heatmap_summary are called with dendrogram=True after a dendrogram has been computed with scanpy.tl.dendrogram, Scanpy issues a dendrogram data not found warning, and re-runs scanpy.tl.dendrogram with default settings (which, in turn, results in scanpy.tl.pca being run).

This is because Scanpy is, by default, looking for adata.uns[f"dendrogram_{groupby}"], but this has not been copied to tmp_adata which infercnvpy.pl.chromosome_heatmap and infercnvpy.pl.chromosome_heatmap_summary pass to scanpy.pl.heatmap.

The problem can be fixed by simply copying adata.uns[f"dendrogram_{groupby}"], if it exists, to tmp_adata.uns, as is already the case with adata.uns[f"{groupby}_colors"]. I'd be happy to make a pull request if that's helpful.

Version information

anndata 0.9.2
scanpy 1.9.4
infercnvpy 0.4.2

[Gregor Sturm]

Hi @jpintar,

a PR would be most welcome!

[Jura Pintar]

PR submitted. Looking at the code again, I realised you'd already fixed chromosome_heatmap with PR #72. I just applied the same fix to chromosome_heatmap_summary and removed a test in both that is rendered moot by the fix.

I also explicitly set the index type of tmp_obs to str, which will avoid AnnData issuing ImplicitModificationWarning: Transforming to str index when tmp_adata is created.

Thanks for the quick response!

[Gregor Sturm]

Closed via #110