Managers 03June09

Agenda/chair: Jane, Midori, Jen, Pascale, David, Judy.

Agenda: Chris

Minutes: Jen

Action items from previous calls

All are carried over from March 25

• Pascale and Jane: Organize conference call on usage of tricky GO terms; include ontology editors and annotators (both ref gen and broader annotation list). -> We'll try to do something for the June Reference genome conference call, June 9th.

• Pascale, Chris: Finish and circulate gp2protein file documentation. IN PROGRESS

• PIs will discuss Jen's funding situation and get back to Jen immediately. (Current contract finishes 14th July.)

• Jen: Paper on the trigger file, send a plan to GO-Top: DONE

• Jane: Ask GO-tops about getting a logo professionally designed for GO. (background: Amelia has suggested that we get a logo professionally designed for GO. She suggested this website: http://99designs.com/. Basically you put up a brief for the logo and an amount of prize money (between 100 and 600 USD) and people submit designs. After a week you choose one and the winner gets the prize money.)
  • We've now got a graphic designer at EBI to do this for us at no cost to GO. We're putting a spec together for him now, will report when we have something.

Action items from today

Action: talk to Chris while he is visiting EBI next week about putting XPs and has_part into write file.

Action: Pascale will write to GO list to arrange the call on binding.

Action: Jane will work with Pascale on sending a survey to GOhelp addresses. Maintain the list and give people the option to come off the list. Initial mail will invite people to join so they are initially opting in.

Discussion items

Binding:

Discussed open calls, but is this for the users, or internally? Perhaps discuss users' analyses? This would be different from actual annotation issues though.

Could have open calls for the two issues separately. Perhaps people could even just listen in, though Jane points out that having a large number of people may make it difficult to get agreement. Judy futher mentions that we need to have the community on board for the system to be useful.

Binding Survey: Too complicated? Lots of options.

Some options were missed out and it is still very complex.

What do the users want from binding? Do people mostly use process? No, they also use function and cell component. Clustering often uses function.

How much do the users actually depend on the binding terms? Do they expect the terms to be exhaustive? For clustering we must surely need an exhaustive graph?

How many users just use the GO to figure out what their gene does? If there are five or six annotations then this can be hard. If they only use for clustering then this is different. We should poll the users on how they use the GO and what they expect. Then we can tell them about the options on how binding could be captured.

GO-Friends is not ideal for communicating with users, as it is mostly tool developers. How do we reach researchers? Maybe a survey via AmiGO? This has been vetoed as too annoying.

Maybe Ruth and Peter need to host the discussion a bit more. The questions are very open and there are only a small number of responses. The last question is 'do nothing' and gets about half of the votes. Many people feel it would be less work to keep things as they are, since there is no really good solution.

Other resources are available for mining the binding of substrates. Peter has a proposal and it is quite good. He proposes only to annotate where there is only binding and no associated catalytic activity.

Pascale has seen 'water binding', but there is water in every protein. It does not make sense to capture this. Also heparin sulphate binding does not make sense to capture. It is used for purification and people annotate it incorrectly.

Judy suggests quality control systems to capture such problems? Perhaps have a couple of people responsible for that. We need an efficient annotation process but some mistakes will always get through. Better to have a checking mechanism.

Pascale: Some of the terms can certainly be confusing. Note which are often misused and set up a system to catch them.

Interpro: all kinase have atp binding and so has an atp binding domain. If we get rid of this then we lose a lot of information.

Maybe for a bunch of terms we could have a set of caveats for annotators to note. It will give a heads-up for users on the pitfalls of the more complex parts of the system.

This would mean not removing catalytic domain binding. It would contravene our guidelines, but maybe then the guidelines need fixed.

In the survey 14% say we should remove all binding terms, and 14% say that we should just remove those with catalytic domain binding.

There is a call coming up about this topic. Peter and Ruth will moderate the call. It will be for the whole consortium.

GO users: maybe the people who write to GOhelp would be the people to ask. We could keep them as a list that we could contact. Perhaps send them surveys.

Can we get their email addresses? Perhaps parse the archive and remove duplicates. Write and request feedback. Also say 'DO you mind being contacted in future' so we are not just spamming people. Make sure we don't send a lot of mails. Just carefully written questions infrequently. Also have a link to the wiki with further information. We should not expose them to the long survey as it is an extremely complex issue and requires a lot of background knowledge.

Would be good to go ahead and discuss this issue on the annotation call. Good to maintain momentum.

Action: Pascale will write to GO list to arrange the call on binding.

Action: Jane will work with Pascale on sending a survey to GOhelp addresses. Maintain the list and give people the option to come off the list. Initial mail will invite people to join so they are initially opting in.

has_part and regulates XPs

Can we start putting this into the write file?

Does the filtering script already filter all this out?

Currently we need to load XPs and write file and then manually strip intersection tags before commit. The users won't see this so there is no real reason not to.

Perhaps load all the XP files that we have confidence in.

Need to get vetted XP files lined up and ready to go.

Action: talk to Chris while he is visiting EBI next week about putting XPs and has_part into write file.

Free beta testing licence for Clackpoint conference call software

Clackpoint is a piece of sofware that we can use for calls via computer (but not currently on the phone). We can have unlimited minutes and up to 250 connections at a time in return for testing the software. If you would like to use it then please do have a go. Please write to Jennifer Deegan for the password, and to send on information on any bugs. A computer must be running Flash 10 for the software to work.

This is the website for connection to the chat room:

http://clackpoint.com/

---

Back to minutes list