What is the correct ontology_curie (UBERON ID) to pull gtex_tissue metadata for GTEx Skeletal Muscle?

Help & Support
1

Hello,

I am running variant predictions using predict_variant and trying to generate a Sashimi plot for GTEx skeletal muscle tissue.

When I filter by UBERON:0002037 (Cerebellum), the prediction works perfectly, and the metadata correctly populates the gtex_tissue column with Brain_Cerebellum, allowing the Sashimi plot to render.

However, when I try to do the same for skeletal muscle using UBERON:0001134, the prediction runs, but the resulting junction_track.metadata does not contain any GTEx mappings (the gtex_tissue column is blank/missing for those tracks). Because of this, plot_components.Sashimi fails to generate a plot.

My setup:

Python

ontology_terms = ['UBERON:0001134'] # Skeletal muscle tissue

variant_output = dna_model.predict_variant(
    interval=variant_interval,
    variant=variant,
    requested_outputs={
        dna_client.OutputType.RNA_SEQ,
        dna_client.OutputType.CAGE,
        dna_client.OutputType.SPLICE_SITES,
        dna_client.OutputType.SPLICE_SITE_USAGE,
        dna_client.OutputType.SPLICE_JUNCTIONS,
    },
    ontology_terms=ontology_terms,
)

ref_out = variant_output.reference
sashimi_component = plot_components.Sashimi(ref_out.splice_junctions)
junction_track = sashimi_component._junction_track

print("Metadata details:\n", junction_track.metadata)
# Result: Tracks are returned, but 'gtex_tissue' info is missing.

My Question: > What is the correct UBERON ID mapped to the GTEx “Muscle - Skeletal” tissue tracks in this model? Is there a master list of how GTEx tissues map to ontology_curie in the metadata?

Thanks!

2

Hi @Pish,

Welcome to the forum! Answers inline:

What is the correct UBERON ID mapped to the GTEx “Muscle - Skeletal” tissue tracks in this model?

The Muscle-skeletal GTEx ontology id is UBERON:0011907.

Is there a master list of how GTEx tissues map to ontology_curie in the metadata?

We have Navigating data ontologies — AlphaGenome which if you load in (Google Colab) you should be able to filter and search all model metadata.

Hope that helps!

Tom

3

Hi Tom,

That worked perfectly! Changing it to UBERON:0011907 and adding the .filter_by_tissue('Muscle_Skeletal') step pulled the tracks right up, and the Sashimi plot generated exactly as expected.

Thanks so much for the quick fix and for linking the AlphaGenome Colab notebook—that is going to save me a ton of time in the future!

Best,

Pish