MalteDLuecken comments

Results 126 comments of


                                            MalteDLuecken

batch_key of highly_variable_genes not working

I've run into the same issue. I'm expecting it to return the a recarray with the fields that `sc.pp.highly_variable_genes()` would return without `batch_key` set. And additionally with the fields `highly_variable_nbatches`...

batch_key of highly_variable_genes not working

There is a further issue with this version of the function as well. If a batch has 0 variance for multiple genes, then the `_highly_variable_genes_single_batch()` function will not work on...

Fix font stack

I think Arial was set as default as it's required for Nature publications. I set the font to arial for a manuscript manually, and then noticed that it's default already.

Define fail modes for `.T`/`.transpose()`

`adata.raw` is still useful if it has the same dimensions as `adata.X` as it is used for statistical testing by default. Therefore you can use it to start log-normalized data,...

improve sc.pp.neighbors parameter docs

I assume UMAP uses a number of top PCs as input as well, no? And do you know if distances are calculated only on 2 UMAP components, or if more...

improve sc.pp.neighbors parameter docs

So the `method='umap'` has nothing to do with using UMAP dimensionality reduction as a distance metric, but instead only refers to a KNN-graph implementation? It seems to me that making...

improve sc.pp.neighbors parameter docs

Thanks for the explanation! I had no idea this was an important bottleneck. I always assumed all distances are calculated... that shows how little I think about optimization ^^. On...

Consider changing detections of strings from `is_string_dtype`

When Chuck Norris uses R, `stringsAsFactors` defaults to False ;)

Consider changing detections of strings from `is_string_dtype`

This conversation sounds relevant to @davidsebfischer who has been working with TCR data in scanpy.

Reading and writing anndata in chunks

@LisaSikkema and @cdedonno this may be relevant for you guys soon as well.