Pancreas system classification

[cmungall]

See thread here: https://twitter.com/compbiologist/status/775100282489430018

relationship to endo and exo systems here:

we model as pancreas consisting of two separate component: endocrine (islets) and exocrine.

Formally, the entailed relationships are:

• pancreas subClassOf overlaps some exocrine system
• pancreas subClassOf overlaps some endocrine system

(from has-part o part-of -> overlaps)

what if we have a sample that is annotated with just 'pancreas'. There is no entailed part-of relationship to endo or exo systems (but there is of course a link to gland, which could be considered a union of endocrine-exocrine). This is formally correct by our modeling, but is it what is expected?

Perhaps we need annotation guidelines saying annotations need to be more specific. Which part of the pancreas? Was the entire pancreas mashed up and sequenced?

Alternatively the relationships can be less strict. For example, we don't require that ALL the parts of the nervous system have nervous functioning (think vasculature). Yet brain is part of the NS. We could just have two part-of links from P to endo and exo. However, I think it better to be precise, in the ontology and in annotations.

How would @bgeedb handle this @ANiknejad @fbastian? What about MGD @tfhayamizu ?

[tfhayamizu]

RE:"Perhaps we need annotation guidelines saying annotations need to be more specific. Which part of the pancreas? Was the entire pancreas mashed up and sequenced?"

Currently in GXD, among 11649 expression results for pancreas and its substructures, 10658 (91.5%) are annotated directly to "pancreas". Of the latter, 2428 (22.8%) are from blot-type assays; 1010 of these from embryonic stages. Thus, it is very likely that the specimen consisted of the entire pancreas, or at least a poorly defined portion of it. Even for in situ-type assays (RNA in situ hybridization, immunohistochemistry, in situ knock-in reporters), it may not be feasible to identify endocrine vs. exocrine structures, certainly not at some developmental stages, and without cell type markers. In these cases, the nature of the experiment necessitates that the annotation be less specific and, furthermore, direct 'part-of' links to endocrine and exocrine systems would be inappropriate.

[fbastian]

I agree with @tfhayamizu. It's tricky, but I think it's safer to not have the part_of relations to endocrine and exocrine systems.

What do you think @ANiknejad?

[ANiknejad]

In last GTEx Harvesting work instructions, they only ask to target 'mid-portion (not tail)' of the pancreas, and actually as said by @tfhayamizu it is difficult to have specimen more specific than 'poorly defined portion' (for Bgee db we annotated this pancreatic mid-portion on UBERON:0001150 (body of pancreas).) So I agree with @tfhayamizu and @fbastian to say safer not having the part_of relations to endo/exo systems.

[cmungall]

It sounds like twitter user compbiologist and ENCODE (@jahilton) should be getting their GTEx annotations direct from BGee since you have already done such a precise thorough job? Rather than everyone annotating the same data slightly differently. Or is the idea that these annotations are pushed back to GTEx?

[fbastian]

Well, we'll contact the GTEx DCC before our release, but I doubt they will consider Bgee as a reliable source for such a widely used dataset at this point... but indeed we went through all pathologist notes for precise checking of the annotations, and of the "healthy" status, and since @ANiknejad has a very good knowledge of Uberon, I believe our annotations are trustworthy.

[dosumis]

I've been playing with using GTEX to test the ability to find tissue signature by grouping GO annotations using Uberon/CL. Works very well for CL. Would have been useful to have an existing mapping to Uberon/CL.

[fbastian]

@dosumis: have you heard about our new tissue signature enrichment tool? :D http://bgee.org/?page=top_anat

About mapping to Uberon: GTEx does provide a mapping to Uberon, through dbGAP, isn't this mapping public?

[gouttegd]

WARNING: This issue has been automatically closed because it has not been updated in more than 3 years. Please re-open it if you still need this to be addressed addressed addressed – we are now getting some resources to deal with such issues.