Difference between revisions of "Managers 04June08"

From GO Wiki
Jump to: navigation, search
 
(24 intermediate revisions by 5 users not shown)
Line 1: Line 1:
 +
[[Category:GO_Managers_Meetings]]
 
GO Managers, Weds. June 4th, 2008 8 AM PDT, 9 AM MDT, 10 AM CDT, 11 AM EDT, 4 PM BST<br>
 
GO Managers, Weds. June 4th, 2008 8 AM PDT, 9 AM MDT, 10 AM CDT, 11 AM EDT, 4 PM BST<br>
  
 
Agenda/chair: <br>
 
Agenda/chair: <br>
  
Minutes:  
+
Minutes: David
  
 
Present:  
 
Present:  
 +
 +
Jen
 +
Midori
 +
David
 +
Pascale
 +
Jane
 +
Mike
 +
  
 
== Action items from last meeting ==
 
== Action items from last meeting ==
  
'''Managers (especially call chair):''' Meeting agenda should be available on Monday (48 hours in advance)
+
'''Managers (especially call chair):''' Meeting agenda should be available on Monday (48 hours in advance)DONE
  
'''Managers:''' Post progress reports on Wiki with links back and forth (e.g. from call agenda to report); refer to wiki during call
+
'''Managers:''' Post progress reports on Wiki with links back and forth (e.g. from call agenda to report); refer to wiki during call DONE
 +
*[[RefGenProgress_2008-06-04]]
  
'''Managers (especially call chair):''' ‘Hot topics and Concerns’ should be at the top of the agenda
+
'''Managers (especially call chair):''' ‘Hot topics and Concerns’ should be at the top of the agenda DONE
  
'''Managers (especially call chair):''' The chair is empowered to manage the call; "Chair trumps PI".
+
'''Managers (especially call chair):''' The chair is empowered to manage the call; "Chair trumps PI".DONE
  
'''Managers:''' Digressions become an action item, not to consume the time of the manager’s meeting
+
'''Managers:''' Digressions become an action item, not to consume the time of the manager’s meeting DONE
  
'''All:''' GO-paid staff should check with PIs before writing papers or going to meetings.
+
'''All:''' GO-paid staff should check with PIs before writing papers or going to meetings.DONE
  
'''GO-top:''' Discuss meeting frequency, who should come, etc.
+
'''GO-top:''' Discuss meeting frequency, who should come, etc.DONE
  
 
'''GO-top:''' Discuss Jane’s letter about review for Human Genetics
 
'''GO-top:''' Discuss Jane’s letter about review for Human Genetics
  
'''Jen:'''Put together bigger picture of function – process task (Jen is on holiday this week but will do this for the next meeting.)
+
'''Jen:'''Put together bigger picture of function – process task SEE BELOW
  
 
==Agenda items==
 
==Agenda items==
  
 
===Hot Topics and Concerns===
 
===Hot Topics and Concerns===
 +
* Montreal GO Meeting (Pascale):
 +
Found a nice place for Oct 21-23 (21, 22: GOC; 23: GO SAB). We could extend 1-2 days for an annotation meeting?
 +
I don't need to confirm this until some time in July but it may be easier to reserve dates now, even if we cancel later.
 +
 +
METING COMMENTS:
 +
 +
Are we going to tack another meeting onto Montreal?
 +
Discuss virtual meeting vs. travel in general
 +
1) Annotation camp: Pascale wants to add another day to the meeting. David is concerned that this will make the meeting too long and people will be burned out. For annotation we need a lot of people to be present.
 +
2) Documentation Meeting?: Maybe only a few people needed.
 +
Midori, the face-to-face meetings are nice, but sometimes we just can’t do that. In these cases the virtual meetings are the next best thing.
  
  
Line 40: Line 61:
 
http://www.life.uiuc.edu/govindjee/photoweb/art/electron-transfer.jpg
 
http://www.life.uiuc.edu/govindjee/photoweb/art/electron-transfer.jpg
  
Problem: The same process has different constituent functions when it happens in different taxonomic groups. For example photosynthesis in chloroplast-containing organisms and cyanobacteria is a bit different from photosynthesis in other photosynthetic organisms (e.g. purple bacteria).
+
'''Problem:'''
 +
*The same process has different constituent functions when it happens in different taxonomic groups.  
 +
*For example photosynthesis in chloroplast-containing organisms and cyanobacteria is a bit different from photosynthesis in other photosynthetic organisms (e.g. purple bacteria).
 +
 
 +
 
 +
'''Proposal 1'''
  
 
As I understand it Chris has an idea of how we could manage still to make the mf-bp links without putting taxon information into the actual ontology file. I think his idea is to make a separate file that would contain the mf-bp links and the taxon constraint information. I guess it would be something like:
 
As I understand it Chris has an idea of how we could manage still to make the mf-bp links without putting taxon information into the actual ontology file. I think his idea is to make a separate file that would contain the mf-bp links and the taxon constraint information. I guess it would be something like:
 +
 +
[Judy:  has_part means 'every instance of A has_part B' doesn't mean that every instance of B has_part A. If you restrict this by taxon then it seems you pervert the meaning of 'has_part'.  Solution? is that parental types have to be defined...have to create, then, 3(or more) different terms to cover the variants, if you will... Once you say 'only in this organism, then when you find in another organisms, it's 'broken'.  This is why and how we got rid of sensu yes?]
 +
 +
(Chris: we're actually perverting the first column, not the relation definition. See my comments below)
  
 
{| border="1" cellpadding="5" cellspacing="0"
 
{| border="1" cellpadding="5" cellspacing="0"
Line 61: Line 91:
  
 
|}
 
|}
 +
 +
 +
'''Proposal 2'''
  
 
His alternative idea is to  subclass the process term  using non-taxon differentia as we do now with the old sensu terms. For example "chloroplast-based  photosynthetic electron transport" however, I can't see how this could be done across the board without making really unusable clunky names. This does not seem a viable option to me.  
 
His alternative idea is to  subclass the process term  using non-taxon differentia as we do now with the old sensu terms. For example "chloroplast-based  photosynthetic electron transport" however, I can't see how this could be done across the board without making really unusable clunky names. This does not seem a viable option to me.  
  
I would like to hear people's views.
+
(Chris: this proposal is entirely compatible with Proposal 1. Proposal 1 is really a first pass at Proposal 2 that avoids having to create clunky subclasses in GO at the early prototypical stage. Once we have collected a sufficient body of links in the tabular format above we can go through en masse and assign subclasses. Or extend OBO-Edit and OBO format to allow qualified links)
 +
 
 +
--
 +
 
 +
'''Proposal 3'''
 +
(Pascale)
 +
Maybe we could figure out a way to capture that at the level of annotation. If there was a way to link F and P annotations, that link itself could be an alternative to the relationship we are looking for between function and process.  
 +
 
 +
 
 +
For example:
 +
A g-protein coupled receptor of Dicty (cAR1, cAMP receptor 1) is annotated as follows
 +
 
 +
 
  
Two further questions:
+
*cAMP binding [F]
 +
*cAMP receptor activity  [F]
  
*Can we make these kinds of generalization when scientists have not checked all taxonomic subgroups?
 
  
* Are we prepared to take on the amount of work that would be involved in situations where there may be many many different variations in different taxonomic groupings.
+
*G-protein coupled receptor protein signaling pathway [P]
 +
*aggregation involved in sorocarp development [P]
 +
*chemotaxis [P]
 +
*activation of adenylate cyclase activity [P]
 +
*positive regulation of protein kinase activity [P]
 +
*regulation of Ras protein signal transduction [P]
 +
 
 +
 
 +
which actually represents several pathways :
 +
-PKA -> myb protein activation -> gene expression -> cell differentiation<br>
 +
-PI3K activation -> PIP3 signaling -> chemotaxis<br>
 +
-PI3K also activates ACA and phosphodiesterase which regulates the response (adaptation, etc)<br>
 +
 
 +
Right now all the annotations can do is list those events in a 'flat' manner (we dont know what process links to other process; ie where the pathway splits and which event is upstream or downstream. ) We could represent the complexity in the ontology as Jen proposed above. Alternatively, if somehow the annotations would allow to see how each process and subprocess and function links to each other using 'annotation relationships' (I made up some possible relationships here)  <br>
 +
 
 +
For example cAR1 annotations could say 
 +
 
 +
*[meta-annotation 1]<br>
 +
cAR1's molecular functions "cAMP binding", "cAMP receptor activity" <br>
 +
<i>activates </i> "G-protein signaling, coupled to cAMP nucleotide second messenger"<br>
 +
<i>activates </i> "positive regulation of phosphoinositide 3-kinase cascade" <br>
 +
which <i>causes </i> "activation of adenylate cyclase activity " <br>
 +
which <i>causes </i> "regulation of phosphidiesterase activity" <br>
 +
which <i>causes </i> "adaptation of signaling pathway"<br>
 +
 
 +
 
 +
*[meta-annotation 2]<br>
 +
cAR1's molecular functions "cAMP binding", "cAMP receptor activity" <br>
 +
<i>activates </i> "G-protein signaling, coupled to cAMP nucleotide second messenger"<br>
 +
<i>activates </i> "positive regulation of phosphoinositide 3-kinase cascade" <br>
 +
which <i>causes </i> "chemotaxis"<br>
 +
 
 +
*[meta-annotation 3]<br>
 +
cAR1's molecular functions "cAMP binding", "cAMP receptor activity" <br>
 +
<i>activates </i> "G-protein signaling, coupled to cAMP nucleotide second messenger"<br>
 +
<i>activates </i> "positive regulation of protein kinase activity" <br>
 +
which <i>causes </i> "regulation of gene expression"<br>
 +
which <i>causes </i> "cell differentiation"<br>
 +
 
 +
The result of this is that one would know that the gene expression/cell differentiation response is mediated through PKA, and the adaptation
 +
is mediated through ACA. If we compared those events in a few organisms we might be able to draw some general principles and infer annotations
 +
based on that ('infered from event observed in xx different taxa').
  
*Are we instead proposing only to cover a small number of well known and researched species, and are we sure that the information is fully available in those cases?
+
(Chris: see [[Annotation_Cross_Products]]. Again this is not opposed to 2 or 1 - pre and post composition are complementary. To support the pathway style annotations above the annotation cross-products would have to be extended. For example, in the above the temporal ordering of events seems significant)
  
or as Chris says:
+
'''Feedback'''
  
- how much time should we spend on each process term in GO?
+
I would like to hear people's views generally. The scientists that I have been working with are concerned, as am I, that this project is just infeasible. We would like to discuss the issues frankly before we use any more time on the pilot project.  
- how important is it to get each individual function step as the  process is instantiated in a given cell? i.e. are gaps a problem
 
- how important is it to get broad coverage across different  organisms or variations of a process? i.e are "representatives" enough?
 
  
It may be that we are just not able to make the bp-mf links because of the complexity involved, but I think we have now reached a place where it makes sense to stop and discuss it before using more time on the pilot project.
+
'''Specific questions'''
  
 +
*Can we make these kinds of generalization when scientists have not checked all taxonomic subgroups?
  
 +
* Are we prepared to take on the amount of work that would be involved in situations where there may be many many different variations in different taxonomic groupings.
  
 +
*Are we instead proposing only to cover a small number of well known and researched species, and are we sure that the information is fully available in those cases?
  
 +
or as Chris says:
  
 +
* how much time should we spend on each process term in GO?
 +
* how important is it to get each individual function step as the  process is instantiated in a given cell? i.e. are gaps a problem
 +
* how important is it to get broad coverage across different  organisms or variations of a process? i.e are "representatives" enough?
  
 +
'''Summary'''
  
 +
It may be that we are just not able to make the bp-mf links because of the complexity involved, but I think we have now reached a place where it makes sense to stop and discuss it before using more time on the pilot project. All thoughts much appreciated.
  
 +
MEETING COMMENTS:
  
 +
Comments on Jen’s proposal for F-P links
 +
Jane’s idea use has_part relationship. Delineate the processes by using has_part. Only put links between the common function. Problem is some taxa will be missing functions. Pascale will write a proposal about linking MF and BP in annotations. Jane will also write a proposal and add it to the wiki.
  
 +
==Summary of RO meeting==
 +
2 things relevant to GO
 +
1) OBO wants a firmer more axiomatic def of regulates. Tanya, Chris and David will work on this.
 +
2) We will begin to promote the idea that the relationship between a gene product and the
 +
molecular function is that the gene product has the potential to execute the function and
 +
the function ontology represents the execution.
  
  
  
===Progress of note===
+
==Progress of note==
  
 
==Next call==
 
==Next call==

Latest revision as of 12:49, 27 June 2014

GO Managers, Weds. June 4th, 2008 8 AM PDT, 9 AM MDT, 10 AM CDT, 11 AM EDT, 4 PM BST

Agenda/chair:

Minutes: David

Present:

Jen Midori David Pascale Jane Mike


Action items from last meeting

Managers (especially call chair): Meeting agenda should be available on Monday (48 hours in advance)DONE

Managers: Post progress reports on Wiki with links back and forth (e.g. from call agenda to report); refer to wiki during call DONE

Managers (especially call chair): ‘Hot topics and Concerns’ should be at the top of the agenda DONE

Managers (especially call chair): The chair is empowered to manage the call; "Chair trumps PI".DONE

Managers: Digressions become an action item, not to consume the time of the manager’s meeting DONE

All: GO-paid staff should check with PIs before writing papers or going to meetings.DONE

GO-top: Discuss meeting frequency, who should come, etc.DONE

GO-top: Discuss Jane’s letter about review for Human Genetics

Jen:Put together bigger picture of function – process task SEE BELOW

Agenda items

Hot Topics and Concerns

  • Montreal GO Meeting (Pascale):

Found a nice place for Oct 21-23 (21, 22: GOC; 23: GO SAB). We could extend 1-2 days for an annotation meeting? I don't need to confirm this until some time in July but it may be easier to reserve dates now, even if we cancel later.

METING COMMENTS:

Are we going to tack another meeting onto Montreal? Discuss virtual meeting vs. travel in general 1) Annotation camp: Pascale wants to add another day to the meeting. David is concerned that this will make the meeting too long and people will be burned out. For annotation we need a lot of people to be present. 2) Documentation Meeting?: Maybe only a few people needed. Midori, the face-to-face meetings are nice, but sometimes we just can’t do that. In these cases the virtual meetings are the next best thing.


mf-bp links

Report on current status by Jennifer Deegan.

The biological process "photosynthetic electron transport" is to be connected to all its consistuent molecular functions in the function ontology. http://www.life.uiuc.edu/govindjee/photoweb/art/electron-transfer.jpg

Problem:

  • The same process has different constituent functions when it happens in different taxonomic groups.
  • For example photosynthesis in chloroplast-containing organisms and cyanobacteria is a bit different from photosynthesis in other photosynthetic organisms (e.g. purple bacteria).


Proposal 1

As I understand it Chris has an idea of how we could manage still to make the mf-bp links without putting taxon information into the actual ontology file. I think his idea is to make a separate file that would contain the mf-bp links and the taxon constraint information. I guess it would be something like:

[Judy: has_part means 'every instance of A has_part B' doesn't mean that every instance of B has_part A. If you restrict this by taxon then it seems you pervert the meaning of 'has_part'. Solution? is that parental types have to be defined...have to create, then, 3(or more) different terms to cover the variants, if you will... Once you say 'only in this organism, then when you find in another organisms, it's 'broken'. This is why and how we got rid of sensu yes?]

(Chris: we're actually perverting the first column, not the relation definition. See my comments below)

Process relationship function taxon restriction
photosynthetic electron transport has_part electron transport, transferring from P700 (photosystem I) to Ferrodoxin Sulphur protein in organisms with chloroplasts (Viridiplantae) and cyanobacteria
photosynthetic electron transport has_part electron transport, transferring from P680 (photosystem II) to plastoquinone in organisms with chloroplasts (Viridiplantae) and cyanobacteria
photosynthetic electron transport has_part electron transport, transferring from cytochrome complex to plastocyanin in organisms with chloroplasts (Viridiplantae) and cyanobacteria
photosynthetic electron transport has_part electron transport, transferring from plastoquinone to cytochrome b6/f in organisms with chloroplasts (Viridiplantae) and cyanobacteria
photosynthetic electron transport has_part electron transport, transferring from plastocyanin to photosystem I in organisms with chloroplasts (Viridiplantae) and cyanobacteria


Proposal 2

His alternative idea is to subclass the process term using non-taxon differentia as we do now with the old sensu terms. For example "chloroplast-based photosynthetic electron transport" however, I can't see how this could be done across the board without making really unusable clunky names. This does not seem a viable option to me.

(Chris: this proposal is entirely compatible with Proposal 1. Proposal 1 is really a first pass at Proposal 2 that avoids having to create clunky subclasses in GO at the early prototypical stage. Once we have collected a sufficient body of links in the tabular format above we can go through en masse and assign subclasses. Or extend OBO-Edit and OBO format to allow qualified links)

--

Proposal 3 (Pascale) Maybe we could figure out a way to capture that at the level of annotation. If there was a way to link F and P annotations, that link itself could be an alternative to the relationship we are looking for between function and process.


For example: A g-protein coupled receptor of Dicty (cAR1, cAMP receptor 1) is annotated as follows


  • cAMP binding [F]
  • cAMP receptor activity [F]


  • G-protein coupled receptor protein signaling pathway [P]
  • aggregation involved in sorocarp development [P]
  • chemotaxis [P]
  • activation of adenylate cyclase activity [P]
  • positive regulation of protein kinase activity [P]
  • regulation of Ras protein signal transduction [P]


which actually represents several pathways : -PKA -> myb protein activation -> gene expression -> cell differentiation
-PI3K activation -> PIP3 signaling -> chemotaxis
-PI3K also activates ACA and phosphodiesterase which regulates the response (adaptation, etc)

Right now all the annotations can do is list those events in a 'flat' manner (we dont know what process links to other process; ie where the pathway splits and which event is upstream or downstream. ) We could represent the complexity in the ontology as Jen proposed above. Alternatively, if somehow the annotations would allow to see how each process and subprocess and function links to each other using 'annotation relationships' (I made up some possible relationships here)

For example cAR1 annotations could say

  • [meta-annotation 1]

cAR1's molecular functions "cAMP binding", "cAMP receptor activity"
activates "G-protein signaling, coupled to cAMP nucleotide second messenger"
activates "positive regulation of phosphoinositide 3-kinase cascade"
which causes "activation of adenylate cyclase activity "
which causes "regulation of phosphidiesterase activity"
which causes "adaptation of signaling pathway"


  • [meta-annotation 2]

cAR1's molecular functions "cAMP binding", "cAMP receptor activity"
activates "G-protein signaling, coupled to cAMP nucleotide second messenger"
activates "positive regulation of phosphoinositide 3-kinase cascade"
which causes "chemotaxis"

  • [meta-annotation 3]

cAR1's molecular functions "cAMP binding", "cAMP receptor activity"
activates "G-protein signaling, coupled to cAMP nucleotide second messenger"
activates "positive regulation of protein kinase activity"
which causes "regulation of gene expression"
which causes "cell differentiation"

The result of this is that one would know that the gene expression/cell differentiation response is mediated through PKA, and the adaptation is mediated through ACA. If we compared those events in a few organisms we might be able to draw some general principles and infer annotations based on that ('infered from event observed in xx different taxa').

(Chris: see Annotation_Cross_Products. Again this is not opposed to 2 or 1 - pre and post composition are complementary. To support the pathway style annotations above the annotation cross-products would have to be extended. For example, in the above the temporal ordering of events seems significant)

Feedback

I would like to hear people's views generally. The scientists that I have been working with are concerned, as am I, that this project is just infeasible. We would like to discuss the issues frankly before we use any more time on the pilot project.

Specific questions

  • Can we make these kinds of generalization when scientists have not checked all taxonomic subgroups?
  • Are we prepared to take on the amount of work that would be involved in situations where there may be many many different variations in different taxonomic groupings.
  • Are we instead proposing only to cover a small number of well known and researched species, and are we sure that the information is fully available in those cases?

or as Chris says:

  • how much time should we spend on each process term in GO?
  • how important is it to get each individual function step as the process is instantiated in a given cell? i.e. are gaps a problem
  • how important is it to get broad coverage across different organisms or variations of a process? i.e are "representatives" enough?

Summary

It may be that we are just not able to make the bp-mf links because of the complexity involved, but I think we have now reached a place where it makes sense to stop and discuss it before using more time on the pilot project. All thoughts much appreciated.

MEETING COMMENTS:

Comments on Jen’s proposal for F-P links Jane’s idea use has_part relationship. Delineate the processes by using has_part. Only put links between the common function. Problem is some taxa will be missing functions. Pascale will write a proposal about linking MF and BP in annotations. Jane will also write a proposal and add it to the wiki.

Summary of RO meeting

2 things relevant to GO 
1) OBO wants a firmer more axiomatic def of regulates. Tanya, Chris and David will work on this.
2) We will begin to promote the idea that the relationship between a gene product and the 
molecular function is that the gene product has the potential to execute the function and 
the function ontology represents the execution.


Progress of note

Next call

June 18, 2008, 8 AM PDT/11 AM EDT/4 PM BST

Agenda: ; Minutes:

Summary of Action items

Back to minutes list