CJM EBI Visit September 2008
Participants: Chris Mungall, Jane Lomax, Jennifer Deegan, Midori Harris.
Q: Ask SAB who would use the MF-BP links. (We know that some of the bacterial sequencing centers are interested.)
- To simplify our initial efforts, only cellular level processes should be linked so (e.g.) no linking of development to transcription factors.
Q: should we do the generic textbook version of processes, or should we do the minimal required set of functions that work in all species?
- DNA binding to transcription is easy. do that sort.
Reactome file: we need reactome BP-MF file and Reactome-to-GO mapping so we can learn from their links. Chris says he could produce this.
- Metacyc has lists linking both function and complex to process by species. Very cool. Good web access.
BFO - basic formal ontology
Process is a term in the BFO, so we can link all process terms in the process ontology to the term 'process' in the BFO. Chris has also proposed that we map GO molecular function to BFO process.
In BFO the term function is defined differently from GO functions. Our functions are mostly single step processes, so would really be BFO processes rather than BFO functions. The BFO uses "function" in the formal ontological sense, meaning the potential to carry out an action. (This is somewhat analogous to a GO annotation, in which we say that a gene product has the ability to carry out a molecular function or participate in a biological process.)
Chris (?) suggested that the ontology group take this proposal to the managers' group and then just do it. We don't need to bother the consortium meeting. The mapping will help allay confusion among ontologists, by making the points that GO MF is an ontology of processes in the BFO sense, and that we have no plans to recapitulate the entire MF ontology in BP.
As Chris noted in his initial email, we have the option of mapping BFO function to something more specific than the root in GO MF, but we decided to stick with the more general mapping for now.
Action: Chris will send mapping file to Midori and David to check. (The "mapping" will consist of a single line stating that GO:0003674 ! molecular_function is_a BFO:function.)
OBO-Edit 2 Release
We discussed whether OBO-Edit 2 is likely to be released before the consortium meeting. Our sense is that it's unrealistic to think OE2 will be released by the meeting, but the end of the year is a possibility (Midori guessed 3-6 months).
We thought it might be feasible to aim for "super beta" by the meeting, by which we mean that all features are fully testable.
OBO-Edit cross products
Add checkbox to parent editor so we can say that a relationship is a cross product or not.
Also request a display element to show whether a relationship is a cross product or not.
- Update: David Sutherland has submitted a feature request (2116808).
- Intersection editor is just a display of the cross product file. Maybe we don't need it to be usable as an editor by release?
- Intersection editor will be removed from the menus for the release. The code will be kept in case it's needed.
- Too much whitespace in intersection editor.
- Press return in "discriminate relationship" box -> and a new one is made. Please turn this off.
- Remove the expression line from the display of intersection editor. It was a debugging tool.
Have to select a number of terms to get several to show up in matrix editor.
- ctrl-click in matrix editor not working.
- assert is_a means automatically put in Is_a hierarchy.
- click on a cell and do 'make' to create the cross products.
- feature request: when we click on a term in a cell please could it show up in the OTE?
To test XP files:
- load write file +
- on wiki: page called:XP:biological_process_xp_cell
- has import file and it's all taking stuff from scratch.
- feature request: Could we select the genus and differentium and have it populate?
We should suppress this component for the release as we will just be using it on the command line.
- This tool makes intersection relationships for the existing regulation terms. They show up in the cross product tab of the text editor. It does not place any terms in the graph so new terms do not show up in the OTE.
- read wiki to find out what's going on.
New Term queue:
- People request regulation terms by submitting the term that is being regulated.
- Goes in a file and then Chris pushes a button and makes the terms.
This file contains the name of the term to be generated and the new term id from our own number range. We enter the numbers from our own ranges so there is no risk of id overlap.
We should start playing with this immediately. Only for regulation terms.
- Update: David and Tanya have done one test run, and are giving Chris feedback.
Q: Can we have this for files in scratch?
- The script will be set up so we can tell it the path of the input GO file and the path of the output file Also the list of terms file.
- Afterwards we have to use obomerge to merge the file back in.
- "Assert implied links" and "remove redundant links" are in the right-click menu.
- Run "assert implied links" first as asserting implied links will make some redundancies.
- Link filter implied links menu option is not working.
- Dotted line means implied but not realised. Looks just like the normal arrow but is very slightly dotted.
- We don't currently have a way to easily assert such links on an individual basis.
In SO these links are never asserted but they would need to be for GO because the reasoner would explode.
If you do 'assert all' and then 'save' then all the intersection links would need to be stripped out. But this doesn't need to be done in OBO-Edit.
- GO curators can add terms to go/scratch/new-term-queue.txt, and Chris' script will run to put the terms in place.
- The reasoned, filtered file will be put in scratch by Chris automatically every night and we should merge in the asserted links with obomerge.
- Filtered means that the intersections will have been stripped out.
- Reasoned here means that the implied links have been asserted.
- Until we have the process-function links in the GO we will need to do a separate file that strips out the BP-MF links.
- Output file is in scratch at go/scratch/regulates_xp_live/go_reglive_withPosNeg_noPF-reasoned-filtered.obo
- At a later date we will use the PF version when we have got the BP-MF links in the live file.
- We should co-ordinate with David and Tanya on when we will start.
- Maybe wait until after Jane merges in PAMGO file before we start doing this, because otherwise OBOMerge will explode. Should be a couple of weeks.
- Bug: We added regulation of fertilization and the reasoning step didn't seem to work, and it took ages. Adding new regulation terms with cross product definitions doesn't yet work.
- Bug report: Feature lost from 1.101 the undo menu doesn't say what it would be undoing.
- Update from David & Tanya: script doesn't fill in definitions. Definitely (pardon the pun) a desirable feature!
- Harder to parse PAMGO regulation terms
- Uses obol instead of the java semantic parser.
- Import file biological_process_xp_multi_organism_process-imports.obo; also name of wiki page (minus "-imports.obo")
This is the import file we need to load in OBO-Edit to see what obol has done to the PAMGO terms.
Maybe easiest to import normal GO and just get PAMGO stuff right first. Doing that now...
There are a large number of terms that don't parse and this might be because the modular bits of the terms aren't represented as individual terms.
Some paths are missing as Jane has not finished the manual part. Hopefully once the manual paths are done obol can do the regulation terms.
One problem seen is that one term is 'formation of X' and another is 'X formation'. Probably best to add a synonym with the words in the right order.
Jane will come back and look at the signaling-related terms after the signaling rearrangement has been committed.
Load live file + cellular_component_xp_self.obo.
Loaded in GO with reasoner on. (Takes ages)
Annotation-Checking Trigger file
Modify link filter to have minimal stanza with just goid and name and relationship required.
We could also load up the annotation list that came from goose that shows where a GO term has only one annotation from a given species.