Summary of todays call and set up next call

[Chris Stoeckert]

Suggested next call - all times EDT. James, please shout if these
times are too late for you.
May 4 - 12 PM, 1PM
May 11 - 12 PM, 1PM
May 12 - 1PM

Attending today: James, Melanie, Alan, Jennifer, Jie, Chris

Terms covered:
1. data transformation parameter (specification)
- agreement that entities like the K in K nearest neighbors are
information content entities and an input to a data transformation.
- disagreement on whether these types of entities are just another
data input or are instructions (specifications).
Action item: OK to add terms under ICE with input to data
transformation as restriction. James will investigate additional
restrictions.

2. genome sequence version (label)
- agreement that this is a label and that the proposed draft
restriction was appropriate
Action item: Chris will come up with needed terms (e.g, assembled
genome sequence) to make this work

3. tree model
- agreement that there is a data representation model of the type tree.
Action item: Chris will incorporate discussion to definition

Thanks,
Chris

[Melanie Courtot]

I am away until the 10th.
I could do 11th at 12PM.
Would May 12 11PM (before the relation call) be an option? 1pm is
generally a bit late for me.

Thanks ,
Melanie

[Fostel, Jennifer (NIH/NIEHS) [C]]

may 4 and 11 work for me, either 12 or 1

[Bjoern Peters]

May 4, 12 pm
May 11, 12 pm
May 12, 11 pm

should work for me.

--
Bjoern Peters
Assistant Member
La Jolla Institute for Allergy and Immunology
9420 Athena Circle
La Jolla, CA 92037, USA
Tel: 858/752-6914
Fax: 858/752-6987
http://www.liai.org/pages/faculty-peters

[James Malone]

May 11th is fine for me, but I'm away all the rest (I'm away all of
May except for 4 days so you guys should go ahead whenever suits you
as my availability will be sketchy at best).

James

[Chris Stoeckert]

Sounds like May 11th at 12PM EDT works best for everyone so far. I've
put it in the google calendar.
Thanks,
Chris

[James Malone]

Hi Chris,

I am actually going to be travelling monday evening now so will not be
able to attend the call as planned. Apologies.

James

[Chris Stoeckert]

Hi James,
Thanks for letting me know. I'll try to send out an email to frame the
discussion so if there are points you can make in response that would
be great.
Safe travels!
Chris

[Chris Stoeckert]

Hi James,

If you get a chance to think about this could you send in any restrictions or differentia for parameter?

Action item: OK to add terms under ICE with input to data  

transformation as restriction. James will investigate additional  

restrictions. 

I know that in bringing this to the DT branch, you proposed parameter as a role. My view is that we should not delay having parameter types in OBI for that discussion on whether roles (realizable entities) can only have independent continuants as bearers. 

Thanks

Chris

On May 5, 2009, at 10:37 AM, James Malone wrote:

[James Malone]

Hi Chris,

I bounced the differences off a few different people and had a few
different answers. I think one of the biggest issues is that is it a
social issue? I can see the argument that parameters, like all data,
are input to DTs, however the difference is that certain parts of
those DTs are considered to be parameters and not the data. This is
what has led me to think parameter could be best modelled as a role.
I can certainly add some under ICE but actually the issue would still
stand and it would still mean I can't answer the competency question
"what are the parameters for k-means?". Alan had some suggestion
about how we might do this via specification the other day, I can't
recall what it was now. Presently I favour role for ICE, but I also
like the idea of a relation has_input_parameter as a subtype of
has_input. I do definitely agree with you though, we shouldn't delay
parameter waiting for roles for ICE to be green lighted - to that end
I would favour just doing it since Barry has already agreed in
principle to this under BFO IIRC (and there are no ontological
restrictions preventing this in OBI).

Thanks!

James

[Alan Ruttenberg]

On Fri, May 8, 2009 at 8:01 AM, James Malone <james....@gmail.com> wrote:

Hi Chris,

I bounced the differences off a few different people and had a few
different answers.  I think one of the biggest issues is that is it a
social issue?  I can see the argument that parameters, like all data,
are input to DTs, however the difference is that certain parts of
those DTs are considered to be parameters and not the data.  

Thought we had discussed and agreed that whether or not something was a parameter was a related to the objective rather than the dt. Here is seems you are back to having them intrinsic to the dts.

 

This is
what has led me to think parameter could be best modelled as a role.
I can certainly add some under ICE but actually the issue would still
stand and it would still mean I can't answer the competency question
"what are the parameters for k-means?".  

Again, inconsistent with the idea that parameterness is related to objectives. A question that acknowledges that would be either one at the instance level - what were parameters for this dt. Or what are parameters for this type of dt as it is used in this kind of part of investigation.

[James Malone]

I initially considered that they are perhaps part of the objective
specification, i.e. they are specified before a particular process is
undertaken and are therein defined as a DT with x parameters and y
input data. However, I can also see that some people consider these a
social convention, this DT always has x as parameters and y as input
data and that doesn't change as it is intrinsic to the definition of
the algorithm. So as far as I can see it depends on your school of
thought. I have received different definitions from different people
- though they did both differentiate between input data and
parameters, they just didn't always agree.

[Alan Ruttenberg]

2009/5/11 James Malone <james....@gmail.com>

I initially considered that they are perhaps part of the objective
specification, i.e. they are specified before a particular process is
undertaken and are therein defined as a DT with x parameters and y
input data.  However, I can also see that some people consider these a
social convention, this DT always has x as parameters and y as input
data and that doesn't change as it is intrinsic to the definition of
the algorithm.  So as far as I can see it depends on your school of
thought.  I have received different definitions from different people
- though they did both differentiate between input data and
parameters, they just didn't always agree.

A good sign that there is still work to do. Have you considered the option of using community specific labels, and using the word "parameter" on an input x community basis?

-Alan

[Alan Ruttenberg]

On Mon, May 11, 2009 at 10:40 AM, James Malone <mal...@ebi.ac.uk> wrote:

Yes, an annotation property would certainly be one way of doing this, though as you know we lose some of our querying power this way.

Depends on the tool you use for querying. It would be helpful to have an example that shows us losing. 

-Alan

 

Alan Ruttenberg wrote:

2009/5/11 James Malone <james....@gmail.com <mailto:james....@gmail.com>>

   I initially considered that they are perhaps part of the objective
   specification, i.e. they are specified before a particular process is
   undertaken and are therein defined as a DT with x parameters and y
   input data.  However, I can also see that some people consider these a
   social convention, this DT always has x as parameters and y as input
   data and that doesn't change as it is intrinsic to the definition of
   the algorithm.  So as far as I can see it depends on your school of
   thought.  I have received different definitions from different people
   - though they did both differentiate between input data and
   parameters, they just didn't always agree.


A good sign that there is still work to do. Have you considered the option of using community specific labels, and using the word "parameter" on an input x community basis?

-Alan

 




   On Mon, May 11, 2009 at 3:00 PM, Alan Ruttenberg

   <alanrut...@gmail.com <mailto:alanrut...@gmail.com>> wrote:
   >
   >
   > On Fri, May 8, 2009 at 8:01 AM, James Malone

   >> > <stoe...@pcbi.upenn.edu <mailto:stoe...@pcbi.upenn.edu>> wrote:
   >> >
   >> > Sounds like May 11th at 12PM EDT works best for everyone so
   far. I've
   >> > put it
   >> >
   >> > in the google calendar.
   >> >
   >> > Thanks,
   >> >
   >> > Chris
   >> >
   >> > On Apr 29, 2009, at 4:04 AM, James Malone wrote:
   >> >
   >> >
   >> > May 11th is fine for me, but I'm away all the rest (I'm away
   all of
   >> >
   >> > May except for 4 days so you guys should go ahead whenever
   suits you
   >> >
   >> > as my availability will be sketchy at best).
   >> >
   >> > James
   >> >
   >> >
   >> >
   >> > On Tue, Apr 28, 2009 at 11:16 PM, Bjoern Peters

   <bpe...@liai.org <mailto:bpe...@liai.org>>

   >> > wrote:
   >> >
   >> > May 4, 12 pm
   >> >
   >> > May 11, 12 pm
   >> >
   >> > May 12, 11 pm
   >> >
   >> > should work for me.
   >> >
   >> > Fostel, Jennifer (NIH/NIEHS) [C] wrote:
   >> >
   >> > may 4 and 11 work for me, either 12 or 1
   >> >
   >> > -----Original Message-----
   >> >
   >> > From: obi-denr...@googlegroups.com
   <mailto:obi-denr...@googlegroups.com>
   >> >

   >> > [mailto:obi-denr...@googlegroups.com

   <mailto:obi-denr...@googlegroups.com>] On Behalf Of Melanie
   Courtot
   >> >
   >> > Sent: Tuesday, April 28, 2009 5:08 PM
   >> >
   >> > To: obi-denr...@googlegroups.com

   <mailto:obi-denr...@googlegroups.com>
   >> >

------------------------------------------------------------------------

------------------------------------------------------------------------------
The NEW KODAK i700 Series Scanners deliver under ANY circumstances! Your
production scanning environment may not be a perfect world - but thanks to
Kodak, there's a perfect scanner to get the job done! With the NEW KODAK i700
Series Scanner you'll get full speed at 300 dpi even with all image processing features enabled. http://p.sf.net/sfu/kodak-com
------------------------------------------------------------------------

_______________________________________________
Obi-datatrfm-branch mailing list
Obi-datat...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obi-datatrfm-branch
 

--
European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, CB10 1SD, United Kingdom
Tel: + 44 (0) 1223 494 676
Fax: + 44 (0) 1223 492 468

[Alan Ruttenberg]

OLS example is helpful. Don't know what their api is. Could investigate. Do they allow queries on annotation properties (I suspect so). Do they allow substring search (I hope so).

Re: consistency, once again, example would be good. What would be an inconsistency case?

-Alan

On Mon, May 11, 2009 at 10:49 AM, James Malone <mal...@ebi.ac.uk> wrote:

Not to mention usual consistency checking of course.


Alan Ruttenberg wrote:

On Mon, May 11, 2009 at 10:40 AM, James Malone <mal...@ebi.ac.uk <mailto:mal...@ebi.ac.uk>> wrote:

   Yes, an annotation property would certainly be one way of doing
   this, though as you know we lose some of our querying power this way.


Depends on the tool you use for querying. It would be helpful to have an example that shows us losing.
-Alan

 


   Alan Ruttenberg wrote:

       2009/5/11 James Malone <james....@gmail.com

       <mailto:james....@gmail.com> <mailto:james....@gmail.com

       <mailto:alanrut...@gmail.com

       <mailto:alanrut...@gmail.com>>> wrote:
          >
          >
          > On Fri, May 8, 2009 at 8:01 AM, James Malone
          <james....@gmail.com <mailto:james....@gmail.com>

       <mailto:james....@gmail.com

       <mailto:stoe...@pcbi.upenn.edu

       <mailto:stoe...@pcbi.upenn.edu

       <mailto:stoe...@pcbi.upenn.edu>>> wrote:
          >> >
          >> > Sounds like May 11th at 12PM EDT works best for
       everyone so
          far. I've
          >> > put it
          >> >
          >> > in the google calendar.
          >> >
          >> > Thanks,
          >> >
          >> > Chris
          >> >
          >> > On Apr 29, 2009, at 4:04 AM, James Malone wrote:
          >> >
          >> >
          >> > May 11th is fine for me, but I'm away all the rest
       (I'm away
          all of
          >> >
          >> > May except for 4 days so you guys should go ahead whenever
          suits you
          >> >
          >> > as my availability will be sketchy at best).
          >> >
          >> > James
          >> >
          >> >
          >> >
          >> > On Tue, Apr 28, 2009 at 11:16 PM, Bjoern Peters
          <bpe...@liai.org <mailto:bpe...@liai.org>

       <mailto:bpe...@liai.org <mailto:bpe...@liai.org>>>

          >> > wrote:
          >> >
          >> > May 4, 12 pm
          >> >
          >> > May 11, 12 pm
          >> >
          >> > May 12, 11 pm
          >> >
          >> > should work for me.
          >> >
          >> > Fostel, Jennifer (NIH/NIEHS) [C] wrote:
          >> >
          >> > may 4 and 11 work for me, either 12 or 1
          >> >
          >> > -----Original Message-----
          >> >
          >> > From: obi-denr...@googlegroups.com
       <mailto:obi-denr...@googlegroups.com>
          <mailto:obi-denr...@googlegroups.com
       <mailto:obi-denr...@googlegroups.com>>
          >> >

          >> > [mailto:obi-denr...@googlegroups.com

       <mailto:obi-denr...@googlegroups.com>
          <mailto:obi-denr...@googlegroups.com
       <mailto:obi-denr...@googlegroups.com>>] On Behalf Of Melanie
          Courtot
          >> >
          >> > Sent: Tuesday, April 28, 2009 5:08 PM
          >> >
          >> > To: obi-denr...@googlegroups.com
       <mailto:obi-denr...@googlegroups.com>

          <mailto:obi-denr...@googlegroups.com
       <mailto:obi-denr...@googlegroups.com>>
          >> >

       <mailto:Obi-datat...@lists.sourceforge.net>

       https://lists.sourceforge.net/lists/listinfo/obi-datatrfm-branch
       


   --    European Bioinformatics Institute, Wellcome Trust Genome Campus,
   Hinxton, Cambridge, CB10 1SD, United Kingdom
   Tel: + 44 (0) 1223 494 676
   Fax: + 44 (0) 1223 492 468

[Alan Ruttenberg]

I had a look at the use case. I'm not sure how having parameter is pivotal to it at all. Seems to me that they want the standard sort of pipeline situation, in which there are inputs of various kinds and outputs of various kinds and some outputs can be set as inputs to something else. 

To the extent that there is a common thread of "the data" running through it, it would part of the protocol to specify what you are interested in. Everything else is only going to be used to hook things up or to be discarded. 

I'll look at the thread about roles for gdc - missed that one.

-Alan

[James Malone]

Was there any outcome from the call last night? Alan, I understand
what you are saying but they still flag certain bits of data as
paramaters. GenePattern use case is one example of this, but most of
the DTs have paramaters, this is a convention from statistics not one
I am defining. It appears we may reach an impass on this. If this is
the case I would propose that I create my own branch file and model
parameters in one of the multiple ways I've suggested. I can show this
at the OBI meeting and we can resolve it then.

Note: I'm at a meeting in Manchester now and will go on vacation
immediately after this until 31st May so my email after the next
couple of days will not be replied to.

James

[Alan Ruttenberg]

I think we should resolve this at the workshop. From my point of view
you haven't addressed the substantive issues - that what are
parameters seems intrinsic to investigations not to dts, that there is
no definition yet offered for parameters, other than that's what they
call them, and that the use case offered to motivate, seems, on
examination not to motivate after all. The OLS use is mentioned but
not elaborated.

At the workshop we should review more than one use case. It would be
helpful if you could consolidate use case and examples so that we can
review them at once. I propose that we might have this be a breakout,
in case everyone isn't interested in it.

-Alan

[James Malone]

Fine.

On Wed, May 13, 2009 at 2:40 PM, Alan Ruttenberg

[Chris Stoeckert]

Hi James,
We decided to defer discussion on parameter. I'll try to get a summary
out today or tomorrow.
Cheers,
Chris

> https://lists.sourceforge.net/lists/listinfo/obi-datatrfm-branch

[Frank Gibson]

James has given several clear use-cases. There has been a definition that has been proposed, re-stated below. There is clear understanding is "some sort of data intput". There has been two alternative solutions proposed

1) is_parameter_of relation child of input_data
2) parameter  is a role.


Please try not to over-engineer the issue, users actually have to use it, so we should be pragmatic and actually represent how it is used, as presented in the use-cases.


A parameter is a role borne by a ICE where the ICE is an input to an information_processing_

process, where the ICE acts as to constrain (settings) the realisation of the algorithm by the information_processing_process.

Frank

--
Frank Gibson, PhD
http://peanutbutter.wordpress.com/

[Alan Ruttenberg]

On Wed, May 13, 2009 at 9:56 AM, Frank Gibson <fgi...@gmail.com> wrote:
> James has given several clear use-cases. There has been a definition that
> has been proposed, re-stated below. There is clear understanding is "some
> sort of data intput". There has been two alternative solutions proposed
>
> 1) is_parameter_of relation child of input_data
> 2) parameter  is a role.
>
>
> Please try not to over-engineer the issue, users actually have to use it, so
> we should be pragmatic and actually represent how it is used, as presented
> in the use-cases.
>
>
> A parameter is a role borne by a ICE where the ICE is an input to an
> information_processing_
> process, where the ICE acts as to constrain (settings) the realisation of
> the algorithm by the information_processing_process.

Hi Frank,
This definition equally applies to the "data" inputs.
The subproperty statement is not a definition.

-Alan