[go] evidence code ontology
Sue Rhee
rhee at acoma.Stanford.EDU
Wed Feb 6 16:50:19 PST 2008
Yes, I understand.
Thanks,
Sue
Smith, Barry wrote:
> I am sure this is already clear to everybody, but please note that
>
> not expressed in
> not located
> is not a constituent of
>
> etc. are not types of relations, any more than non-cats are types of
> cats.
> If we not the X is not located in Y, then we do not represent this as
>
> X stands in the relation of not-located-in to Y
>
> rather, we put:
>
> not (X located-in Y)
>
> BS
>
> At 01:26 PM 2/6/2008, Sue Rhee wrote:
>> Yes, I agree that it would be helpful to make any relationships
>> between ontologies and annotations explicit for many reasons
>> (Hilmar's example of contradicting evidences for the same assertion
>> is an excellent one). The following are relationships between GO and
>> genes or between genes and genes that we have been using in TAIR for
>> ~6 years or so. They probably should go into the RO, if not there
>> already.
>>
>> +-------------+-----------------------------------------------+
>> | annotations | description |
>> +-------------+-----------------------------------------------+
>> | 35665 | involved in |
>> | 563 | functions as |
>> | 7193 | expressed in |
>> | 111 | not involved in |
>> | 181 | not expressed in |
>> | 401 | related to |
>> | 72 | required for |
>> | 6 | not required for |
>> | 29373 | located in |
>> | 23 | not located in |
>> | 222 | is subunit of |
>> | 222 | constituent of |
>> | 215 | regulates |
>> | 10266 | functions in |
>> | 16 | represses |
>> | 90 | is regulated by |
>> | 62 | suppresses gene |
>> | 46 | enhances gene |
>> | 3 | partially enhances gene |
>> | 6 | partially suppresses gene |
>> | 384 | expressed during |
>> | 53 | expressed only in |
>> | 5 | expressed only during |
>> | 80 | is downregulated by |
>> | 29784 | has |
>> | 1237 | has protein-protein physical interaction with |
>> | 4 | binds to cis-element of |
>> | 9 | acts upstream of |
>> | 44 | acts downstream of |
>> | 27 | has protein-DNA interaction with |
>> | 437 | has protein modification of type |
>> | 6 | protein is modified by |
>> | 57 | does not have |
>> | 4 | contributes to |
>> | 35 | colocalizes with |
>> | 1 | does not function in |
>> | 355 | is upregulated by |
>> | 36 | is not regulated by |
>> | 4 | is not a constituent of |
>> | 9 | not expressed during |
>> +-------------+-----------------------------------------------+
>>
>>
>>
>>
>> Larry Hunter wrote:
>>>
>>> Given Sue's comment that
>>>
>>>> Perhaps the 'inferred from' should probably be omitted from term
>>>> names and the ontology renamed as Evidence Source Ontology. In my
>>>> view, it represents types of methods that are used to generate
>>>> results that are used as evidences for an assertion. So eventually
>>>> I see the leaf nodes of this ontology being individual assay types.
>>>> In this view, the ECO could be considered as some sort of meta
>>>> ontology for the OBI.
>>>
>>> it seems clear that the relation(s) between such evidence sources
>>> and the propositions (hypotheses? assertions?) that they inform
>>> still need to be defined, and that these epistemic relations could
>>> be used by the GO, phenoscape and others to make more explicit the
>>> relationship of evidence to their annotations.
>>>
>>> Currently, the GO makes instance level assertions about genes having
>>> a particular molecular function, participating in a particular
>>> biological process, or being located in a particular cellular
>>> component, and links each such assertion to an evidence code
>>> (sometimes specifying the PMID of a publication that contains the
>>> evidence). The instance-level relationship ("has function"
>>> "participates in" "is located in") is implicit, as is the
>>> (meta-)relationship between the particular instance level
>>> relationship and the PMID or other source of evidence.
>>>
>>> There may well be value in making these implicit relationships more
>>> explicit, both for automated reasoning, and for making assertions
>>> about universals. There are already RO terms (or proposed terms)
>>> that will make possible explicit instance level statements (e.g.,
>>> <gene> "participates in" <process>) -- what I am trying to get to is
>>> the set of relations necessary to make the relationship between such
>>> statements and the evidence for them explicit. Since all of the GO
>>> evidence codes (and apparently the ones used by Karp's xCyc systems)
>>> are related to literature assertions, I am inclined to make the
>>> definitions in terms of OBI DENRIEs), but more on that in another
>>> email.
>>>
>>> Larry
>>>
>>>
>>>
>>>>
>>>> Sue
>>>>
>>>> Suzanna Lewis wrote:
>>>>> Chris said it much better than I.
>>>>>
>>>>> I agree fully with this summary.
>>>>>
>>>>> -S
>>>>>
>>>>> On Feb 5, 2008, at 4:01 PM, Chris Mungall wrote:
>>>>>
>>>>>>
>>>>>> At the moment, ECO isn't very ontology-like. As Larry pointed
>>>>>> out, it's more of a terminology of codes with specific
>>>>>> instructions on what fields should be filled in the GAF. The
>>>>>> adoption of ECO by phenoscape and the extension to homology-based
>>>>>> evidence types is a good thing and should hopefully keep ECO
>>>>>> general enough for use outside GO. At the same time, ECO has to
>>>>>> be guaranteed to fulfill certain GO requirements.
>>>>>>
>>>>>> OBI is a massive and impressive effort - but it is still young.
>>>>>> Its emphasis is on representing actual experiments and associated
>>>>>> entities - rather than codes - which is a good thing. It is
>>>>>> certainly the OBO ontology for representing an experiment. But
>>>>>> ECO doesn't represent experiments per se - it represents evidence
>>>>>> and/or inference, which presents additional challenges.
>>>>>>
>>>>>> I would say the evidence group should be actively engaging OBI,
>>>>>> we should be trying to define ECO terms using OBI terms where
>>>>>> possible. Likewise the OBI group should treat ECO as a set of
>>>>>> requirements. And I think we can use OBI to provide a better
>>>>>> structure to ECO. But I think it is too soon to talk of
>>>>>> immediately subsuming ECO into OBI - sorry Sue, you don't get out
>>>>>> of it that easily!!
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Feb 5, 2008, at 3:40 PM, Sue Rhee wrote:
>>>>>>
>>>>>>> Hi GO and Phenoscape colleagues,
>>>>>>>
>>>>>>> Larry Hunter pointed me to the OBI project
>>>>>>> (http://obi.sourceforge.net) recently, which I wasn't aware of.
>>>>>>> It seems to me that the ECO is redundant to and much less
>>>>>>> developed than the OBI and I would like to suggest that OBI take
>>>>>>> over merging ECO into their ontology. Please let me know if you
>>>>>>> have any views on this.
>>>>>>>
>>>>>>> Best,
>>>>>>> Sue
>>>>>>>
>>>>>>> --
>>>>>>> Sue Rhee
>>>>>>> Staff Scientist
>>>>>>> Carnegie Institution, Department of Plant Biology
>>>>>>> 260 Panama Street, Stanford, CA 94305
>>>>>>> Email: (650) 325-1521 x251
>>>>>>> Fax: (650) 325-6857
>>>>>>>
>>>>>>
>>>>
>>>> --
>>>> Sue Rhee
>>>> Staff Scientist
>>>> Carnegie Institution, Department of Plant Biology
>>>> 260 Panama Street, Stanford, CA 94305
>>>> Email: (650) 325-1521 x251
>>>> Fax: (650) 325-6857
>>
>> --
>> Sue Rhee
>> Staff Scientist
>> Carnegie Institution, Department of Plant Biology
>> 260 Panama Street, Stanford, CA 94305
>> Email: (650) 325-1521 x251
>> Fax: (650) 325-6857
>>
>>
>
>
--
Sue Rhee
Staff Scientist
Carnegie Institution, Department of Plant Biology
260 Panama Street, Stanford, CA 94305
Email: (650) 325-1521 x251
Fax: (650) 325-6857
More information about the Go
mailing list