What is an entity

[cannin]

Extract from a discussion with Huaiyu that illustrate the problem (and the fix):

I see. Thanks. This is another example of using multiple ways to represent a pathway, just like the way we do in activity flow. It will be nice to have it in the example of the spec.

Thanks,

Huaiyu

Nicolas Le novère wrote: > > Huaiyu Mi wrote: > > > >
>> >> By the way, here is another question on the ER spec. It states that >> >> each entity can only appear once in the map. I am not sure if define >> >> entity here. I looked in the spec, but could not find. One can argue >> >> that proteins with two different modification states are two different >> >> entities. >> >>
> > > > They could. You can define two entities called Prot-nonphos and Prot-phos. > > Then you create statevariable existence on each with alternative values T > > and F. Then a kinase would stimulate the value T on Prot-phos and stimulate > > the value F on Prot-nonphos. A phosphatase would stimulate the value F on > > Prot-phos and stimulate the value F on Prot-nonphos. > > > > Easier to use an assignment :-) > >

Reported by: lenov

[cannin]

• assigned_to: nobody --> lenov
• status: open --> pending

Original comment by: lenov

[cannin]

• labels: --> specification content
• status: pending --> open

Original comment by: lenov

[cannin]

• status: open --> pending

Original comment by: lenov

[cannin]

What we are really saying here is that each gene product can only be represented once in an ER diagram, right?

Original comment by: huaiyumi

[cannin]

I'd like us to discuss this at COMBINE. I think the fix described above is wrong. This is an example of the problem of dealing with the entity as a class and then dealing with the individual instances of the entity (the reason we have the cis/trans).

Original comment by: stumoodie

[cannin]

Some clarification added in the spec regarding the entity itself. Now the semantic of the entity use in a map is still an ongoing debate, so I let this tracker opened.

Stuart, you are right and wrong I think. What I say is that you can have XP and XnP in an ER map to represent two states. This is the choice of the modeler. But then you loose the fact that XP and XnP represent the same underlying physical reality (you may complement that by annotations). This may look silly for biochemistry - and probably is, I would not do that - but the situation will be different as we move towards representation of physiological models or models used in drug discovery (PK.PD). The situation is totally similar to PD, where we can have an three EPNs representing MAPK with a state variable taking the values " ", "P" and "2P" or three EPNs MAPK, MAPK-P, MAPK-2P. The generator of the map chooses how s/he wants to represent things. The important thing is that it is interpreted the same way, and that s/he is conscious of the expressivity limitations attached to thte choice

Original comment by: lenov

[cannin]

• status: pending --> open

Original comment by: lenov

[cannin]

• status: open --> pending

Original comment by: lenov

[cannin]

• status: pending --> closed

Original comment by: lenov