search

MetabolicAtlas / data-generation

Process the raw data-files for ingestion into the Neo4j database
MIT License
0 stars 0 forks source link

fix: use correct metadata field to populate RNA levels #15

Closed e0 closed 3 years ago

e0 commented 3 years ago

The isHuman from const isHuman = metadataSection.organism === 'Homo sapiens'; is used to populate the HPA RNA levels. In the newer data files, the organism field is missing, which means isHuman is always false, which in turn means the HPA RNA levels do not get populated. This PR uses the short_name field instead, which exists in both the older and newer data files.

e0 commented 3 years ago

In theory, using the organism is more generic, but in practice it does not make a difference for us. I haven't tested this but the change looks good.

It looks like the organism field is missing from the newer data files. For example: https://github.com/MetabolicAtlas/data-files/blob/feat/more-models/integrated-models/Human-GEM/Human-GEM.yml

I assume the feat/more-models branch represents the new format going forward so I made this change. The short_name also exists in both versions so it looks like the safer option for both cases.

Edit: updated PR description to reflect this as well.

inghylt commented 3 years ago

Perhaps silly question, but how can I recreate the problem? I assume it's related to what's shown in Interaction Partners?

e0 commented 3 years ago

Perhaps silly question, but how can I recreate the problem? I assume it's related to what's shown in Interaction Partners?

Sorry, I should have been more specific in the PR description. You could checkout the feat/more-models branch from the data-files repo and rebuild the stack. After this if you try to go to the 3D map viewer and apply some RNA levels, all the circle nodes should be gray.