[Obo-anatomy] UBERON. A way to retrieve relationships?

[Guille Reales]

Dear community,

I have gene expression data from several animals, which I'd like to compare. These data have UBERON identifiers for the tissues in which they're expressed, however, many tissues belong to different (yet, potentially) comparable anatomical entities (e.g. "kidney" and "adult mammalian kidney"). I know that UBERON, just like GO, has a hierarchical structure, so I thought that perhaps I could compare tissues belonging to the same group withinthe UBERON hierarchy. A way to access such data is via OLS, yet I haven't found an easy way to retrieve the UBERON relationships among multiple anatomical entities. I also investigated R package "rols", but I read that it might not be suitable for such task.

So my question is: Is it a easy-ish way to extract the hierarchical relationships between UBERON terms (from a list of UBERON ids, for example)?

Thanks a lot.

Best,

--

GUILLERMO REALES MONTEAGUDO

PhD Student in Genetics and Molecular Biology, PPGBM.

Universidade Federal do Rio Grande do Sul (UFRGS).

Porto Alegre, RS. Brazil.

[Chris Mungall]

Hi Guille

It sounds like you are looking for a programmatic interface, and you might prefer R. The last time I looked I found a number of R generic ontology libraries were problematic. But as I recall there is a bgee library specifically for working with expression data annotated with uberon, you might find this useful.

We have a python library that can be used with Uberon
http://ontobio.readthedocs.org/
(you can use locally downloaded files, or it will transparently wrap a sparql service)

There are a number of REST endpoints that can be used as well. E.g. the OLS API, or the ontobee SPARQL endpoint.

How exactly do you intend to compare? Are you doing this for pairs of genes?

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot_______________________________________________
Obo-anatomy mailing list
Obo-a...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obo-anatomy

[Guille Reales]

Hi Chris,

Actually, I retrieved my data mainly from Bgee, as I have a number of genes I want to compare in terms of levels/amplitude of expression across species. As I learnt in the process (I'm an utter rookie at working with expression and ontology, to be honest), anatomical homology is something to take into account when making such comparisons. Fortunately, each anatomical entity at Bgee is accompanied by its corresponding UBERON id and name, yet there is a different degree of specificity in tissues and organs among species (e.g. humans have information on most subparts of the brain, whereas other species have only "prefrontal cortex", or simply "brain"). I was wondering if there is a way to somehow group anatomical entities (thus, using Uberon hierarchy, for example) so I can increase the number of anatomical entities for comparison. That took me to the quest for a tool to retrieve the hierarchy, or at least the relationships between Uberon terms.

I used the BgeeDB for retrieving expression data, though I am not sure that it is suitable for this specific task. As for the other resources, I will carefully look into them.

Thank you very much for your comments,

Best,

[Chris Mungall]

There are a number of ways to do this but I think I still need a bit more information.

For any pair of genes, you could get the bgee expression values for all classes, and then report values for shared superclasses. E.g. if G1 is expressed in the cerebellum, and G2 in the brain, you can roll up the the level of brain. This part should be straightforward using an ontology library or a pre-calculated table of ancestry relationships

[Frederic Bastian]

Jumping into the discussion a bit late:

In Bgee we propagate the expression data using the Uberon hierarchy, exactly as Chris described. You can safely select a list of terms your are interested in, all expression data from substructures will also be included.

We also maintain a list of homology relations between anatomical structures to perform valid comparisons across animals: https://github.com/BgeeDB/anatomical-similarity-annotations

Regarding the example of aggregating data from different "kidneys" for instance, this might need to be done using "transformation_of" relations in Uberon.

Don't hesitate to contact me if you need more information about how to best use Bgee data.

Cheers,
Frederic.

Le 28.04.18 à 02:10, Chris Mungall a écrit :

[Guille Reales]

Thank you for your comments Frederic, I'll definitely contact you in the next future for more specific questions.

Cheers,