Mapping gene names to Ensembl IDs

[JosephDiPalma]

Did you have a chance to explore mapping the gene names to Ensembl IDs?
I'm looking into that now and don't know if there's a recommended way.

[guillaumejaume]

As of today, we only provide support to de-alias HUGO gene names, https://github.com/mahmoodlab/HEST/blob/main/src/hest/HESTData.py#L1178.

@konst-int-i, do you have a code snipped to provide?

@JosephDiPalma, PR welcome! Handling gene names is always complex.

[JosephDiPalma]

I'll look into it further.
Ultimately, I'd like to use it in something like Geneformer which requires the Ensembl IDs.
Any further ideas would be appreciated too.

If I figure out a good solution, I'll send a PR.

[konst-int-i]

Hi @JosephDiPalma,

I also encountered this issue in the past and here is a working version of a fix that I have not PR'ed yet.

It's handled relatively easily using sc.queries.biomart_annotations. I strongly recommend caching as the initial queries can be slow.

    def _ensembleID_to_gene(adata: sc.AnnData, species: str): 
        """
        Converts ensemble gene IDs to gene names using BioMart annotations
        """
        org = "hsapiens" if species == "Homo sapiens" else "mmusculus"
        
        annotations = sc.queries.biomart_annotations(org=org,attrs=['ensembl_gene_id', 'external_gene_name'], use_cache=True)
        ensembl_to_gene_name = dict(zip(annotations['ensembl_gene_id'], annotations['external_gene_name']))
        adata.var['gene_name'] = adata.var_names.map(ensembl_to_gene_name)
        

        # Filter out genes where the conversion returned NaN       
        adata.var_names = adata.var['gene_name'].fillna('')
        valid_genes = adata.var['gene_name'].notna()
        adata = adata[:, valid_genes]

        return adata  

Let me know if you have any troubles getting this to work - I will also create a draft PR to continue this discussion.