recipe_monocle on kb_python output

Hi Xiaojie!

When running dyn.pp.recipe_monocle on the output from kb_python (for conventional scRNAseq), some input ENSEMBL IDs find duplicate hits for gene symbols, or find no hit, as shown in the error below:

45 input query terms found dup hits:
	[('ENSG00000229425', 2), ('ENSG00000182378', 2), ('ENSG00000178605', 2), ('ENSG00000226179', 2), ('E
192 input query terms found no hit:
	['ENSG00000278198', 'ENSG00000273496', 'ENSG00000278782', 'ENSG00000277761', 'ENSG00000277726', 'ENS
Pass "returnall=True" to return complete lists of duplicate or missing query terms.

To avoid errors from dynamo in subsequent steps, a workaround is to use adata[:, adata.var_names.notnull()]

after running dyn.pp.recipe_monocle. But I was wondering if there is a more permanent solution? It looks like this is an issue of the mygene dependency that contains a different ENSEMBL ID dataset than the one used in kb_python. @Lioscro might have some thoughts too from kb_python?

Thanks, Jorge

[Lioscro]

@jmartinrufino Let’s discuss this on the kb-python repo, since it seems it may be a bug. Here’s the issue: https://github.com/pachterlab/kb_python/issues/160

[Lioscro]

Is working with gene names directly an option? You can provide --gene-names to kb count to have all the vars be gene names (gene names that have multiple gene IDs will be summed).