BioTop
Chris Mungall
I'm not sure if this list is purely for BFO or also for related
efforts; anyway, this relates to the previous discussion on OBO-RO
I just discovered this:
http://www.ifomis.uni-saarland.de/biotop
BioTop is an upper level ontology for biology; the types in BioTop
extend the types in BFO.
I have some specific questions about the content of the ontology -
hope you all don't mind if I air these on the BFO list. Then I have
some questions about where this ontology should fit into the BFO-OBO
ecology which Bill mentioned in his previous email.
Some questions on the ontology:
PopulationOfPrimates is defined as a population of mammals.
Presumably an error?
YeastCell is_a UnicellularOrganism
Fungi is_a MulticellularOrganism
??
This is presumably an error.
In fact I would altogether question the value of having two classes:
MonocellularOrganism and MulticellularOrganism. These classes are not
monophyletic and thus when BioTop is merged with any phylogenetically
sound organismal taxonomy we will end up with multiple inheritance.
Instead I would create two qualities: Unicellular and multicellular,
add them to OBO-PATO[1] as "Qualities of an Organism". These
qualities can then be tied via inherence relations to appropriate
classes in an external organismal taxonomy (thus unifying ontologies
with cladistic approaches quite nicely). This is very much in line
with the approach taken by the Tree of Life ontology working groups.
In fact I'd question having any organism classes in BioTop at all -
there's certainly no need for Rat in a top level ontology. I presume
this was done for illustrative purposes. Perhaps the pedagogic
ontology could be separated from the real BioTop? Or is BioTop
entirely pedagogic?
I'm not sure where the dividing line would be. It may be useful to
have some upper level organismal classifications in there. For
example, there is currently strong demand in OBO for BacterialColony,
but no consensus on where the definition should live. It may be
appropriate to define this in BioTop, as a subclass of
PopulationOfBacteria. However, it is more likely that this class will
be managed as part of a bacterial ontology.
Populations are defined using 'grains'; the definition given for
grainOf is:
Grains are the constitutent parts of Pluralities. They are of the
same sort. Generally pluralities have a high number of grains. The
loss of a grain has no effect on the sortality of the identity of the
plurality.
How do the mereologists feel about this? Do we need grains as well as
parts? Should grainOf be elevated to BFO?
Some questions on future evolution of this ontology:
Do you imagine creating a distinction between canonical continuants
and processes and pathological continuants and processes (along the
lines of OBR)?
Will this ontology be actively maintained? What is the procedure for
getting new classes added or amended?
I see there are plans to link the root nodes of various OBO
ontologies to BioTop (presumably either via subclassOf or
equivalentClass as appropriate). This will be incredibly useful! I
have already started on this effort so it would be good to merge
efforts.
However, I do think that the BFO interpretations of the various OBO
root nodes should not be imposed on the various ontology maintainers
- this will take some discussion. It will probably be easiest and
most useful if you start with the OBO Foundry since these adhere to
more rigorous principles. For example, with PATO[1], which I help
maintain, I can state that the root node "Quality" is entirely
consistent with BFO:Quality. In other cases it's more dubious - for
example, with the Sequence Ontology (SO), as Rob has pointed out
(We're working on formalising SO such that there will be a clearer
definition of the root node).
Cheers
Chris
[1] http://www.bioontology.org/wiki/index.php/PATO:Main_Page
Stefan Schulz
thank you a lot for your valuable feedback to BioTop.
>
> I have some specific questions about the content of the ontology -
> hope you all don't mind if I air these on the BFO list. Then I have
> some questions about where this ontology should fit into the BFO-OBO
> ecology which Bill mentioned in his previous email.
>
> PopulationOfPrimates is defined as a population of mammals.
> Presumably an error?
Yes, of course
> YeastCell is_a UnicellularOrganism
> Fungi is_a MulticellularOrganism
You are right we have to revise this. Fungi (which is a plural form) is
the name of the the kingdom which includes both unicellular (yeast)
and multicellular fungi (e.g. mushrooms). I suggest to substitutes
"Fungi" by
"Multicellular Fungus".
> In fact I would altogether question the value of having two classes:
> MonocellularOrganism and MulticellularOrganism. These classes are not
> monophyletic and thus when BioTop is merged with any phylogenetically
> sound organismal taxonomy we will end up with multiple inheritance.
> Instead I would create two qualities: Unicellular and multicellular,
> add them to OBO-PATO[1] as "Qualities of an Organism". These
> qualities can then be tied via inherence relations to appropriate
> classes in an external organismal taxonomy (thus unifying ontologies
> with cladistic approaches quite nicely). This is very much in line
> with the approach taken by the Tree of Life ontology working groups.
But if you use these qualities for defining unicellular and
multicellular organisms then you also end up with multiple inheritance
if you link to an organismal taxonomy. Personally, I do not find this
necessarily bad provided that the potential polyhierachical parents
are fully defined.
However, what does "unicellular organism" mean ? It is something which
is both an organism and a cell. An a multicellular organism is an
organism which is not a cell.
If I defined it in a sense "An unicellular organism is an organism
which has exactly one cell as proper part", then we had to answer the
question which is the complement to this part.
> In fact I'd question having any organism classes in BioTop at all -
> there's certainly no need for Rat in a top level ontology. I presume
> this was done for illustrative purposes. Perhaps the pedagogic
> ontology could be separated from the real BioTop? Or is BioTop
> entirely pedagogic?
You are right, we should do it.
> I'm not sure where the dividing line would be. It may be useful to
> have some upper level organismal classifications in there. For
> example, there is currently strong demand in OBO for BacterialColony,
> but no consensus on where the definition should live. It may be
> appropriate to define this in BioTop, as a subclass of
> PopulationOfBacteria. However, it is more likely that this class will
> be managed as part of a bacterial ontology.
We can add this. Should we do the same with fungi ? BTW, is there a
clear unity criterion for fungi ? Apart from yeast as an unicellular
fungus, how to define and delimit one multicellular fungus ? I'm not a
biologist, so I need your help.
> Populations are defined using 'grains'; the definition given for
> grainOf is:
> Grains are the constitutent parts of Pluralities. They are of the
> same sort. Generally pluralities have a high number of grains. The
> loss of a grain has no effect on the sortality of the identity of the
> plurality.
>
> How do the mereologists feel about this? Do we need grains as well as
> parts? Should grainOf be elevated to BFO?
>
We used the grainOf relation from a paper in the Journal of Biomedical
Informatics written by Alan Rector and Thomas Bittner. Thomas
presented another paper on this issue at FOIS 2006. We should
definitely have a way to express pluralities, and I think there is a
good consensus to consider the relation between a collection and a
collection "element" as a kind of mereological relation.
> Some questions on future evolution of this ontology:
>
> Do you imagine creating a distinction between canonical continuants
> and processes and pathological continuants and processes (along the
> lines of OBR)?
At the level of BioTOP I would not make this distinction. WRT
processes there are many neutral ones (e.g. cell death, necrosis,
inflammation), the pathological or canonical quality of which depends
on many contextual parameters in an organism.
> Will this ontology be actively maintained? What is the procedure for
> getting new classes added or amended?
Yes. The procedure will be explained on the BioTOP site soon.
> I see there are plans to link the root nodes of various OBO
> ontologies to BioTop (presumably either via subclassOf or
> equivalentClass as appropriate). This will be incredibly useful! I
> have already started on this effort so it would be good to merge
> efforts.
Yes this is the major purpose of BioTOP. But wait for the next release.
> However, I do think that the BFO interpretations of the various OBO
> root nodes should not be imposed on the various ontology maintainers
> - this will take some discussion. It will probably be easiest and
> most useful if you start with the OBO Foundry since these adhere to
> more rigorous principles.
This is, exactly,our plan.
> For example, with PATO[1], which I help
> maintain, I can state that the root node "Quality" is entirely
> consistent with BFO:Quality. In other cases it's more dubious - for
> example, with the Sequence Ontology (SO), as Rob has pointed out
> (We're working on formalising SO such that there will be a clearer
> definition of the root node).
>
Best regards
Stefan
Holger Stenzhorn
I have just set up a proper discussion group (aka mailing list) for
BioTop. It can be found at http://groups.google.com/group/biotop.
Holger