Hinxton working sessions Oct. 2-3, 2006
Hinxton working sessions Oct. 2-3, 2006
David Hill, Jane Lomax, Jen Clark, Midori Harris
1) Used OBO-Edit filters to identify "is_a orphans," i.e. terms that didn't have at least one all-is_a path
2) Went through orphans one by one to find is_a parent(s); created some new grouping terms, and will probably add more (see below)
3) Used OBO-Edit reasoner to detect redundant links and resolved them.
4) Made plans for completing the work via Webex [or equivalent]
a) Jane and Jen will do the first pass to look at the remaining is_a orphans and create relationships for the straightforward ones.
b) Save any problematic ones for a Webex session
c) Also via Webex: for each high-level term, look at children and add grouping terms as needed, esp. if there are a lot of children and/or 'natural' groupings are obvious
More on 3c: For the work during and immediately after the meeting, we decided to add high-level is_a parents for now, and to go back to each high-level term to see whether the children should be grouped further. For example, we stuffed a lot of development terms directly under 'anatomical structure development' (GO:0048856); many could probably be moved further down to be children of new or existing children of GO:0048856.
5) Once Jane/Jen and the rest of the group have made is_a paths for the remaining orphans, we will have a 'round robin' editing session similar to the ones we did for CNS. Each person in the group will work on the ontology for a week, refining children and relationships in an area of the graph. Areas needing attention will be development, metabolism, movement/localization and other areas with large numbers of direct is_a children. At the end of each week, the working graph will be forwarded to Jane and she will merge it with the live GO to keep things in synch.---[dph;10/10/2006]
Also see additions to to-do list.
Rules, rationales, clever ideas
1. Insight from the cell cycle: 'X cycle process'
--cell cycle ----[p] cell cycle process ------[i] cell cycle phase (e.g. M phase) ----[p] cell cycle phase (e.g. M phase)
--cellular process ----[i] cell cycle --[i] cell cycle process
In other words, a cell cycle process (or can think of it as a cell cycle subprocess) is a process that forms part of the cell cycle
Any other cycle can be treated the same way; the parent of 'X cycle process' will be either cellular process (GO:0009987) or multicellular organismal process (GO:0032501), as appropriate. Furthermore, this approach can also be applied to non-cyclical processes!
2. Development, developmental process - handle as for cell cycle, i.e.
--biological process --[i] development ----[p] developmental process --[i] developmental process
There was much discussion of whether the existing 'development' term (GO:0007275) should remain 'development' (i.e. the e complete developmental program of any organisms; the whole process), 'developmental process' (a collection of processes involved in development), or 'multicellular organismal development' (as for 'development', but excluding single-celled organisms. (Note that there was never any doubt that we want all of these terms; the question was which one we already have.) Despite the fact that option 3 (multicellular organismal development) is the name GO:0007275 has in the final IsaComplete.obo file from the meeting, Midori votes for GO:0007275 to be 'development', and for 'developmental process' to get a new ID, because there is less risk of any existing annotations going wrong. There are two S. pombe gene products annotated directly to GO:0007275!
Whichever way we do it, the current children of GO:0007275 will have to be examined to see which ones should go under 'developmental process and which under multicellular organismal development (also see to-do list).
3. Morphogenesis is equivalent to morphogenesis of an anatomical structure, so we don't need two separate terms. note, however, that we do need separate terms for 'development' and 'anatomical structure development' because (a) 'development' includes both morphogenesis and maturation, and (b) development of an anatomical structure can be distinguished from development of an entire organism.
Anatomical structure development is_a developmental process.
Morphogenesis is_a developmental process, and part_of anatomical structure development.
For GO's purposes, define 'anatomical structure' as part of an organism (note: FMA has organism is_a anatomical structure).
4. We'll also have reproduction and reproductive process:
--biological process ----[i] reproduction ------[p] reproductive process ----[i] reproductive process
Much discussion of how cellular reproduction (GO:0032505) should relate to reproductive cellular process (GO:0048610); result was to keep both, refine defs, rename GO:0032505 to 'reproduction of a single-celled organism', and add two children to GO:0048610, 'reproductive process in single-celled organism' and 'reproductive cellular process in multicellular organism'.
5. Metabolism, metabolic process, & transport
We'll do the same again (though it's not done yet) ...
--biological process ----[i] metabolism ------[p] metabolic process ----[i] metabolic process
Note that the existing term, GO:0008152, will become 'metabolic process' because we've used it with essentially that meaning, i.e. a collection of processes.
Now that 'X metabolism' terms are defined as 'The chemical reactions and pathways involving X', perhaps we can make X transport a child of X metabolism ... an idea for further discussion. We would NOT make transport of anything other than X itself (precursors 7 whatnot) a child of X metabolism.
6. We're proposing to obsolete 'establishment of dosage compensation' (GO:0007550) because (a) it's undefined; (b) it's not used for annotations; and, most importantly, (c) it seemed to us that there are several different mechanisms of dosage compensation that do not have an establishment mechanism in common. It would be better to have specific terms for establishment of each kind of dosage compensation that are parts of the respective dosage compensation processes and is_a children of the type of process. For example, establishment of dosage compensation by Barr body formation is a type of heterochromatin formation. Also see to-do list.
7. A handy way to tell if a process term should be is_a or part_of its parent: ask whether an instance of the child is an instance of the entire process? If yes, the child is is_a; but if it's an instance of only a portion of the parent process, the child is part_of.
1. What was the reason for naming GO:0042065 'glial growth' rather than 'glial cell growth'? Can it be renamed and made is_a cell growth?
2. What does 'absorption of light' (GO:0016037) really mean? The current def ('The reception of a photon by a cell.') is not very helpful, and we couldn't think of an Aristotelian one ...
3. Is a hair follicle an organ?
4. Should we have a generic 'regulation' term? (If so, it would have two children - regulation of biological process and regulation of a trait.)
5. Will we want 'cell-cell docking'? How similar are the molecules involved in, say, conjugant formation and [some other example]?
6. Should 'reproduction of a single-celled organism' be under reproduction (at present it is)? Or is it just cell division? Biologists would expect to find single-celled organism reproduction under reproduction!
7. Do we have 'elongation' under 'growth'? [answer: no; the closest thing is unidimensional cell growth, GO:0009826, which is_a cell growth and has 'cell elongation' as a narrow synonym]
8. Can establishment of dosage compensation' (GO:0007550) be defined sensibly, and given an is_a parent, or should we go ahead with the proposed obsoletion? Ask Michael about dosage compensation generally.
8 COMMENT: Also see SourceForge item 1569608, on the component term 'dosage compensation complex'. (midori 10-10-06)
9. What does 'establishment of synaptic specificity at neuromuscular junction' (GO:0007529) really mean? Does it refer to which neurotransmitter is used? Ask Erika.
10. Should we have four types of 'follicular fluid formation in ovarian follicle antrum'? At present GO:0001548 is part_of four different parent processes.
11. Can the 'formation of catalytic spliceosome ...' terms (GO:0000349 & GO:0000350) be renamed to avoid saying 'formation'? The 'formation' wording made us think of protein (or RNP) complex assembly, but the terms describe conformational changes. Also, does the protein folding parentage make sense? Ask Karen C.
12. Does 'formation of generative and vegetative cells' (GO:0009564) mean the formation of a two-celled haploid microgametophyte -- and if so, can it be reworded? And should it have child terms for formation of generative cell and formation of vegetative cell? Jen to check textbooks.
13. Do we need a term for 'protein oxidation'? We have 'lipoprotein oxidation' (GO:0042161), but it's not defined. [Note added during write-up: also have 'protein amino acid oxidation' GO:0018158]