location of GO:0048226 ! Casparian strip

[gocentral]

violates x-disjoint.owl (more on this file at the workshop)

reason: part of extracellular region and part of cell

Reported by: cmungall

Original Ticket: geneontology/ontology-requests/9076

[gocentral]

all of these fall in the same category of error

GO:0005607-laminin-2 complex GO:0009530-primary cell wall GO:0009531-secondary cell wall GO:0020005-symbiont-containing vacuole membrane GO:0020006-symbiont-containing vacuolar membrane network GO:0048196-middle lamella-containing extracellular matrix GO:0048217-pectic matrix GO:0048222-glycoprotein network GO:0048223-hemicellulose network GO:0048224-lignin network GO:0048225-suberin network

Original comment by: cmungall

[gocentral]

The term causing most of the problems is:

middle lamella-containing extracellular matrix ; GO:0048196 is_a: extracellular matrix is_a: external encapsulating structure.

Tanya... can you shed any light on the plant structure of GO:0048196? It's defined as: The matrix external to the cell, composed of the cell wall and middle lamella.

Thanks, Becky

Original comment by: rebeccafoulger

[gocentral]

• assigned_to: nobody --> tairtb

Original comment by: rebeccafoulger

[gocentral]

Reviewing this, I see the comment:

Note that this term does not have 'extracellular region ; GO:0005576' as a parent because in plants the extracellular matrix is considered part of the cell.

However, right now, it does have 'extracellular region ' as a parent. I suggest solving this problem by obsoleting 'extracellular matrix' as this is an artificial grouping term as currently defined, grouping by string and not by location:

A structure lying external to one or more cells, which provides structural support for cells or tissues; may be completely external to the cell (as in animals) or be part of the cell (as in plants).

Instead, we'd just retain the two children terms:

'middle lamella-containing ECM' which would be part of the cell and 'proteinaceous ECM' which would be part of the extracellular region , no more mixing of external to cell and part of cell.

We'd have to disentangle the other ECM terms (binding, assembly, disassembly, regulation,...) to figure out which were plant and which were animal or just create specific ones for each but I think this would serve to clarify the issue, not confuse it.

Thoughts?

Original comment by: tberardini

[gocentral]

I think grouping classes are very useful when there is some shared biological characteristic, be it functional, structural, homology, etc. I think it gets very dubious with terms like "epidermis development" - I would be in favour of splitting that one.

I always thought that ECM, whilst different in constituents in plants and animals, seemed like a reasonable grouping class, but given what you say I think a split might be in order. This would solve the part/not-part issue. The current def doesn't really make sense does it? The implication is that plant ECM is simultaneously outside and inside the cell. It should really say "external to the plasma membrane" (or something)

Seems like a pretty big change though. And what about dicty, fungi..?

Didn't we discuss this at one of the GO meetings? Michael had an analogy about waxing a car (or waxing something).

An alternative to splitting the class is to re-think "extracellular", and perhaps subdivide this - e.g. extrernal to plasma membrane vs external to the cell+its parts.

Original comment by: cmungall

[gocentral]

I was just looking at annotations to 'ECM' and 'proteinaceous ECM'. From that glance, it seems that most of the ECM annotations are for mouse/human proteins and could easily be transferred down to 'PCM'. There's one TAIR annotation and possibly a few Dicty ones.

>The current def doesn't really make sense does it? The implication is that >plant ECM is simultaneously outside and inside the cell. It should really >say "external to the plasma membrane" (or something)

Yes.

If we want a grouping term for the plant and animal ECMs, perhaps ECM needs to be a direct child of cellular component and the children terms point to the appropriate cell part vs. extracellular region part parents.

Original comment by: tberardini

[gocentral]

any chance this can be done before the hinxton workshop? If not I'll remove the disjointness axiom.

Original comment by: cmungall

[gocentral]

• priority: 5 --> 6

Original comment by: cmungall

[gocentral]

Let's remove that disjointness axiom for now. If you like, we can discuss as a case in point in Hinxton. What if this axiom existed? How would we identify the offending terms? Good practice, I think. Maybe we'll even be able to tease out the biology.

Original comment by: tberardini

[gocentral]

I think this has now been addressed by the changes made to ECM. Please reopen if necessary.

Original comment by: tberardini

[gocentral]

• status: open --> closed-fixed

Original comment by: tberardini