Cellular component vs cellular location
[from Chris to ontology-editors list]
I'd like to briefly revisit this issue
We have a request for a term "Fully spanning the plasma membrane" https://sourceforge.net/tracker/index.php?func=detail&aid=2831884&group_id=36855&atid=440764
This a location rather than a cellular component. The determining factor would be whether instances of the type would have mass. A nucleus instance has mass, but it's not clear what a "Fully spanning the plasma membrane" instance is.
This is also the case for the intrinsic/extrinsic terms. See the email from Alan below which I guess never made it onto the annotation list (I guess we should probably set up a public GO ontology discussion list distinct from GO friends to stop people spamming the wrong lists?)
Of course these terms are useful and I'm not suggesting getting rid of current annotations. But it is important to be clear what the terms in CC are. This is especially important as other groups start using GO terms in cross-products.
The options as I see it are:
[1] Interpret all GO CC terms as locations.
Thus GO:0005634 ! nucleus would be interpreted as "located in the nucleus". Note that this is not the current interpretation as far as I see it; an instance of GO:0005634 is an instance of a nucleus. When we have an association between a gene product and a nucleus then we interpret this as the gene product being localized to the nucleus.
[2] Introduce a new high level term "cell location".
The is_a hierarchy would be something (roughly) like
cell location
membrane location
extrinsic
spanning
intrinsic
fully spanning
full spanning plasma membrane
(it would be more complex with dual parentage as we have the cross product between membrane type and spatial qualifier)
There would be additional contained_in/part_of links following the current structure, so query results, enrichment etc would remain roughly the same
[3] Use spatial qualifiers in annotation
Here we would actually obsolete the locational terms, and replace them with annotation qualifiers
* extrinsic, intrinsic: membranes only * overlaps * fully contained by * fully spanning
[4] Keep things as they actually are and not worry about giving a coherent explanation as to what a cell component is.
I am against [1] for reasons I can expand on. I am also against [4] but partially resigned to it. I prefer [3] to [2]
This is also related to the host <CC> terms as well, but I think this is best dealt with separately
Begin forwarded message:
> From: Barry Smith <phismith@buffalo.edu> > Date: August 1, 2009 6:58:05 AM PDT > To: Alan Ruttenberg <alanruttenberg@gmail.com>, Suzanna Lewis <suzi@berkeleybop.org>, bfo-discuss@googlegroups.com > Cc: annotation@genome.stanford.edu, huntley@ebi.ac.uk > Subject: [bfo-discuss] Re: Example of one of the problems in cellular component > Reply-To: bfo-discuss@googlegroups.com > > > >>> >>> >>> ---------- Forwarded message ---------- >>> From: Rachael Huntley <huntley@ebi.ac.uk> >>> Date: Fri, Jul 31, 2009 at 5:50 AM >>> Subject: [Annotation] GPI-anchored proteins >>> To: annotation@genome.stanford.edu >>> >>> >>> Hi all, >>> >>> I'm after some advice. I'm a little confused about these two terms, >>> with respect to GPI-anchored proteins; >>> >>> GO:0031224 intrinsic to membrane - Located in a membrane such that >>> some covalently attached portion of the gene product, for example part >>> of a peptide sequence or some other covalently attached moiety such as >>> a GPI anchor, spans or is embedded in one or both leaflets of the >>> membrane. Note that proteins intrinsic to membranes cannot be removed >>> without disrupting the membrane, e.g. by detergent. >>> >>> GO:0019898 extrinsic to membrane - Loosely bound to one surface of a >>> membrane, but not integrated into the hydrophobic region. Note that >>> proteins extrinsic to membranes can be removed by treatments that do >>> not disrupt the membrane, such as salt solutions. >>> This term can be used instead of these obsolete terms: GO:0015025 >>> GPI-anchored membrane-bound receptor (consider GO:0019898) >>> >>> Both mention GPI anchor, the first (intrinsic to membrane) in the >>> definition and the second as a suggestion to use extrinsic to membrane >>> instead of the obsolete GO:0015025 GPI-anchored membrane-bound >>> receptor >>> >>> I don't know much about GPI-anchored proteins, but from what I can >>> gather they can be extracted by detergent-solubilizing a membrane >>> (PMID:19374451) which would suggest use of the term GO:0031224 >>> intrinsic to membrane. However, the GPI-anchor can be disrupted by >>> phospholipase C, thus releasing the associated protein, which would >>> suggest use of the term GO:0019898 extrinsic to membrane. >>> >>> Additionally, GO:0031224 intrinsic to membrane has the child >>> GO:0031225 anchored to membrane (Def: Tethered to a membrane by a >>> covalently attached anchor, such as a lipid moiety, that is embedded >>> in the membrane. When used to describe a protein, indicates that none >>> of the peptide sequence is embedded in the membrane.) which would be a >>> term I would use for GPI-anchored proteins. >>> >>> Can anyone suggest whether GPI-anchored proteins should be annotated >>> to extrinsic or intrinsic to membrane. Either way, it looks as though >>> the ontology could be refined in this area. >>> >>> Thanks for your help. >>> >>> Rachael. >>> >>> -- >>> GOA and IntAct Curator >>> European Bioinformatics Institute >>> Welcome Trust Genome Campus >>> Hinxton >>> Cambridge, CB10 1SD >>> UK >>> >>> Tel: 01223 492515 >>> Fax: 01223 494468 > >> At 09:27 AM 7/31/2009, Alan Ruttenberg wrote: >> Nice discussion. I'll note, however, that an adjective can't be a >> place. Taj Mahal, versus Tah Mahalish. Or "above the Taj Mahal" (a >> location) versus "floating above the Taj Mahal" (a relation to a >> location). >> >> Cellular components are either places that things can be located in, >> or substances that are part_of cells. >> And GO:0031225: anchored to membrane >> >> Is neither of these - it is a state of affairs or disposition or something. >> >> -Alan > > I agree with Alan that there are cellular component terms which need > cleaning up. However, I note that, according to BFO, objects can have > both other objects and also holes (cavities) as parts. Thus for > instance your gut and your nostrils are parts of you. This is one of > the reasons why it is wrong to see organisms as sums of molecules, for example. > BS > BS >