specimen etc.

[Bjoern Peters]

For the manuscript, wee need to resolve the specimen / sample etc.
modeling. It comes up in every reasonable example we have to illustrate
the use of OBI. Once calls are restarted, I think that should be the
primary focus. The approach taken of just doing the 'tissue sample from
organism' (or something like that) first will be fine.

- Bjoern

--
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

[Frank Gibson]

On Wed, May 20, 2009 at 9:34 PM, Bjoern Peters <bpe...@liai.org> wrote:

For the manuscript, wee need to resolve the specimen / sample etc.
modeling. It comes up in every reasonable example we have to illustrate
the use of OBI. Once calls are restarted, I think that should be the
primary focus. The approach taken of just doing the 'tissue sample from
organism' (or something like that) first will be fine.

One solution is documented here
http://docs.google.com/Doc?id=dhgmt5k6_2025vhhj4c3&hl=en_GB

Frank



 

- Bjoern

--
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

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

[Melanie Courtot]

Frank,

I don't seem to be able to edit the doc, so here are a few comments:

1. http://groups.google.com/group/obi-biomaterial/browse_thread/thread/375775d40909a7ee?hl=en

As you can see in the notes from the call we propose a definition for sample role

2. if your arms fall off they don't fit our definition, as they are not output of taking sample from organism. For that reason, your arms would indeed not be EOO/specimen, and we need to tie the definition of EOO/specimen to the process.

We talked about this quite a bit in our latest calls, and as Bjoern mentions we plan to discuss this further during our next call, would there be an option for you to attend? It would be much easier for all of us if we were to discuss together :)

Melanie

---

Mélanie Courtot

TFL- BCCRC

675 West 10th Avenue

Vancouver, BC

V5Z 1L3, Canada

[Frank Gibson]

On Mon, May 25, 2009 at 8:53 AM, Melanie Courtot <mcou...@gmail.com> wrote:

Frank,

I don't seem to be able to edit the doc, so here are a few comments:

1. http://groups.google.com/group/obi-biomaterial/browse_thread/thread/375775d40909a7ee?hl=en

As you can see in the notes from the call we propose a definition for sample role

This is frustrating. Why did you do this? There has been a definition of sample proposed for over a year and a half in the issue tracker and also restated in the google doc. Did you not read it?   What you have come up with is circular -  sampling produces a sample, a sample is the output of sampling - what is a sample? I thought I described this succinctly in the document.




 

2. if your arms fall off they don't fit our definition, as they are not output of taking sample from organism. For that reason, your arms would indeed not be EOO/specimen, and we need to tie the definition of EOO/specimen to the process.

I am not sure what you are implying with you falling arms suggestion. Please look at the definition of sample in the google doc. You are confusing producing a sample - which has a specific meaning of a representative of a whole with  creating an "input"

A sample has a specific meaning -  you can not accept or reject you null hypothesis unless your thing you are investigating is a representative of a population (a class) - this is basic statistics.


 

We talked about this quite a bit in our latest calls, and as Bjoern mentions we plan to discuss this further during our next call, would there be an option for you to attend? It would be much easier for all of us if we were to discuss together :)

I will try, but can you specifically state where you disagree with the proposal - prove is wrong rather that just coming up with different circular definitions, I can not follow your reasoning when this happens and therefore I can not respond contructively.

Ultimately I think you are confusing "producing an input" with "producing a sample". The definition of "sample" is extremely clear - all of statistics are based on it. You can not accept or reject your null hypothesis unless you have a representative sample of the population (class)

Frank

 

[Ryan Brinkman]

It seems one of the problems here is a sample in the biological sense
(e.g., a bit of tissue ripped from my liver) from sample in the
statistical test (a random selection of people on the bus asked if they
support a fare increase). Would a compound term label be helpful here
"statistical_sample" vs. "biomaterial_sample". While definitions of each
will make the distinction obvious, the term "sample" is overloaded, and
this may help lessen confusion among users.

-rb

[Frank Gibson]

On Mon, May 25, 2009 at 6:11 PM, Ryan Brinkman <rbri...@bccrc.ca> wrote:

It seems one of the problems here is a sample in the biological sense
(e.g., a bit of tissue ripped from my liver) from sample in the

Why is a bit of tissue ripped from your liver? because its assumed to be representative of the whole liver (population/class)

 

statistical test (a random selection of people on the bus asked if they
support a fare increase).

Its exactly the same. The sample, or the statistical_sample is a representative (or assumed representative) of the whole/population/class). I do an experiment on some tissue ripped from you  liver because I assume the sample will be representative of your whole liver, and perform as your liver or demonstrate some infection that I can infer your whole liver has.

 

Would a compound term label be helpful here
"statistical_sample" vs. "biomaterial_sample". While definitions of each
will make the distinction obvious, the term "sample" is overloaded, and
this may help lessen confusion among users.

I have never carried out an experiment on some material that I did not believe (or assumed) to represented a larger population or class. I would suggest that the majority of scientific reporting concludes "that we observe this trait in this material, therefore we assume all material show this trait" - you cant do this unless you material you observe is representative of a whole.

The fact that the original proposal was that sample was a role ( now realizable entity) allows an material to be representative or not depending if it bears the role, therefore accounting for your compound option - there is no such thing as biological :)


Frank

 

[Melanie Courtot]

On 25-May-09, at 2:19 AM, Frank Gibson wrote:

On Mon, May 25, 2009 at 8:53 AM, Melanie Courtot <mcou...@gmail.com> wrote:

Frank,

I don't seem to be able to edit the doc, so here are a few comments:

1. http://groups.google.com/group/obi-biomaterial/browse_thread/thread/375775d40909a7ee?hl=en

As you can see in the notes from the call we propose a definition for sample role

This is frustrating. Why did you do this? There has been a definition of sample proposed for over a year and a half in the issue tracker and also restated in the google doc. Did you not read it?   What you have come up with is circular -  sampling produces a sample, a sample is the output of sampling - what is a sample? I thought I described this succinctly in the document.

We did this as follow up to numerous discussions that took place during the calls.

If you were to consider the explanation and notes in the email thread I pointed to earlier, you would also notice the following note: " this is a design decision: we talk about sample of organismal

origin for now, knowing that later on we will need to deal with statistical sample, environmental sample and expand on this"

I fail to see the circularity you mention.

Proposed definition of sample of organism role = realized in process

where x is proxy for o and x has part o' or derives from o' and o' is

specified output of taking sample from organism and o is specified

input

example: liver slice: human specified input of taking sample, output

is liver, liver slice is derived from liver 

Your definition: A sample is a role that is borne by a material entity that is, the part (subset) of a population (whole or class) which is assumed to be representative of the population.

Our definition encompass the representative aspect you mention (is proxy for) and precise it by tying it to the process.

In order to move forward the biomaterial branch decided to address use cases on an individual basis and answer requirements formulated by our users.  I personally don't see the issue if this were to require a change in the ontology.

Please also bear in mind that this is work in progress, discussed during the calls, and still may evolve.

 

2. if your arms fall off they don't fit our definition, as they are not output of taking sample from organism. For that reason, your arms would indeed not be EOO/specimen, and we need to tie the definition of EOO/specimen to the process.

I am not sure what you are implying with you falling arms suggestion. Please look at the definition of sample in the google doc. You are confusing producing a sample - which has a specific meaning of a representative of a whole with  creating an "input"

The falling arms suggestion was yours, taken from your document: "Using the relation is_derived_from if my arm falls off it is still derived from me I dont beleive it suddenly also becomes a specimen."

I was pointing out that I agree (yay!!!) with you: indeed your falling arms are no specimen. Being a specimen involves an active process. If your arms fall off there has not been realization of a process with the intent of your arms becoming specimen. 

Counter example, if a leaf falls from a tree and I collect it with the intent of using it as a specimen then it indeed is a specimen.

A sample has a specific meaning -  you can not accept or reject you null hypothesis unless your thing you are investigating is a representative of a population (a class) - this is basic statistics.


 

We talked about this quite a bit in our latest calls, and as Bjoern mentions we plan to discuss this further during our next call, would there be an option for you to attend? It would be much easier for all of us if we were to discuss together :)

I will try, but can you specifically state where you disagree with the proposal - prove is wrong rather that just coming up with different circular definitions, I can not follow your reasoning when this happens and therefore I can not respond contructively.

Ultimately I think you are confusing "producing an input" with "producing a sample". The definition of "sample" is extremely clear - all of statistics are based on it. You can not accept or reject your null hypothesis unless you have a representative sample of the population (class)

Again, see above. We make some design decision during the calls in order to solve this issue. We never succeeded to have agreement between all concerned parties about defining things at high level, so we are taking the bottom up approach and addressing one thing at a time. 

I hope this addresses some of your concerns. If you are interested in going further with this discussion please consider attending the calls: if it helps we could try and arrange for a more suitable time. Hopefully this will help you follow our reasoning and decrease frustration on both sides :)

Cheers,

Melanie

Frank

 

Melanie

On 21-May-09, at 1:43 AM, Frank Gibson wrote:

On Wed, May 20, 2009 at 9:34 PM, Bjoern Peters <bpe...@liai.org> wrote:

For the manuscript, wee need to resolve the specimen / sample etc.
modeling. It comes up in every reasonable example we have to illustrate
the use of OBI. Once calls are restarted, I think that should be the
primary focus. The approach taken of just doing the 'tissue sample from
organism' (or something like that) first will be fine.

One solution is documented here
http://docs.google.com/Doc?id=dhgmt5k6_2025vhhj4c3&hl=en_GB

Frank



 

- Bjoern

[Ryan Brinkman]

I don’t disagree with you. I was wondering if there might be a way to avoid confusion by biologists who might confuse EOO/specimen with sample as the use the words interchangeably, perhaps by using another label for “sample”.

[Frank Gibson]

On Mon, May 25, 2009 at 7:11 PM, Melanie Courtot <mcou...@gmail.com> wrote:

On 25-May-09, at 2:19 AM, Frank Gibson wrote:

On Mon, May 25, 2009 at 8:53 AM, Melanie Courtot <mcou...@gmail.com> wrote:

Frank,

I don't seem to be able to edit the doc, so here are a few comments:

1. http://groups.google.com/group/obi-biomaterial/browse_thread/thread/375775d40909a7ee?hl=en

As you can see in the notes from the call we propose a definition for sample role

This is frustrating. Why did you do this? There has been a definition of sample proposed for over a year and a half in the issue tracker and also restated in the google doc. Did you not read it?   What you have come up with is circular -  sampling produces a sample, a sample is the output of sampling - what is a sample? I thought I described this succinctly in the document.

We did this as follow up to numerous discussions that took place during the calls.

If you were to consider the explanation and notes in the email thread I pointed to earlier, you would also notice the following note: " this is a design decision: we talk about sample of organismal

origin for now, knowing that later on we will need to deal with statistical sample, environmental sample and expand on this"

I fail to see the circularity you mention.

Proposed definition of sample of organism role = realized in process

where x is proxy for o and x has part o' or derives from o' and o' is

specified output of taking sample from organism and o is specified

input

example: liver slice: human specified input of taking sample, output

is liver, liver slice is derived from liver 

Your definition: A sample is a role that is borne by a material entity that is, the part (subset) of a population (whole or class) which is assumed to be representative of the population.

Our definition encompass the representative aspect you mention (is proxy for) and precise it by tying it to the process.

I think your are taken the meaning of proxy to its limits. Why did you feel the need to generalise the original definition? What part of the original definition was wrong to motivate you to create another definition?. In addition the original definition is a sample of material, your are taking about an organism sample, which I would argue that this conflates an organism that is a sample of a population of organism and a part of an organism that representative of a part of an organism. A confusion that is accounted for in the original definition.

Meaning that and organism_sample and a tissue_sample are defined classes bearing the role sample

A proxy means acts_as and I would suggest it does not mean the same as is_subset_of or is_representative_of. However, I can see the train of thought that as something is_subset_of  then it can act_as the larger whole, in a loose sense of the definition. Following this line of thought I would still argue that you need the statement that it is an representative of the class - then and only then can you infer that it is a proxy for, if you insist on including this statement, which I don't believe is neccessary.


 

In order to move forward the biomaterial branch decided to address use cases on an individual basis and answer requirements formulated by our users.  I personally don't see the issue if this were to require a change in the ontology.

Which use-case proves that the examples in the google document do not work?


 

Please also bear in mind that this is work in progress, discussed during the calls, and still may evolve.

I appreciate this, please also bear in mind that this was originally submitted over a year and half ago, it is documented with examples and as yet has not been proven to be wrong, but is still not in the ontology file

 

2. if your arms fall off they don't fit our definition, as they are not output of taking sample from organism. For that reason, your arms would indeed not be EOO/specimen, and we need to tie the definition of EOO/specimen to the process.

I am not sure what you are implying with you falling arms suggestion. Please look at the definition of sample in the google doc. You are confusing producing a sample - which has a specific meaning of a representative of a whole with  creating an "input"

The falling arms suggestion was yours, taken from your document: "Using the relation is_derived_from if my arm falls off it is still derived from me I dont beleive it suddenly also becomes a specimen."

I was pointing out that I agree (yay!!!) with you: indeed your falling arms are no specimen. Being a specimen involves an active process. If your arms fall off there has not been realization of a process with the intent of your arms becoming specimen. 

yes, this process, assuming that you mean sample, is the result of a sampling process, which has the objective to produce an output that is a subset of, or representative of the input.

 

Counter example, if a leaf falls from a tree and I collect it with the intent of using it as a specimen then it indeed is a specimen.

If by "using it as a specimen" you mean, you assume to mean the fallen leaf is representative of all leaves of the tree then this is still a sampling process. Whether you use it in another experiment or not is irrelevant in this case. Samples are the result of sampling process. An input to a process can also be a sample, not all inputs are samples



 

A sample has a specific meaning -  you can not accept or reject you null hypothesis unless your thing you are investigating is a representative of a population (a class) - this is basic statistics.


 

We talked about this quite a bit in our latest calls, and as Bjoern mentions we plan to discuss this further during our next call, would there be an option for you to attend? It would be much easier for all of us if we were to discuss together :)

I will try, but can you specifically state where you disagree with the proposal - prove is wrong rather that just coming up with different circular definitions, I can not follow your reasoning when this happens and therefore I can not respond contructively.

Ultimately I think you are confusing "producing an input" with "producing a sample". The definition of "sample" is extremely clear - all of statistics are based on it. You can not accept or reject your null hypothesis unless you have a representative sample of the population (class)

Again, see above. We make some design decision during the calls in order to solve this issue. We never succeeded to have agreement between all concerned parties about defining things at high level, so we are taking the bottom up approach and addressing one thing at a time. 

What is the issue you are trying to solve? What is specifically wrong with the proposal according to your case-study you are trying to model?
 

I hope this addresses some of your concerns.

My concern is that you have not proved the proposal does not work, yet you seem to be ignoring it without stating why it does not work for the case study your are trying to model

 

If you are interested in going further with this discussion please consider attending the calls: if it helps we could try and arrange for a more suitable time. Hopefully this will help you follow our reasoning and decrease frustration on both sides :)

I think a start would be a statement showing the case-study that does not fit the proposal and then I can take it from there. I am finding it difficult to make time to attend the calls. This is why I took the time to actually present a proposal, although I have receive no feedback on the proposal showing it does not work.


Cheers

Frank


 

Cheers,

Melanie

Frank

 

Melanie

On 21-May-09, at 1:43 AM, Frank Gibson wrote:

On Wed, May 20, 2009 at 9:34 PM, Bjoern Peters <bpe...@liai.org> wrote:

For the manuscript, wee need to resolve the specimen / sample etc.
modeling. It comes up in every reasonable example we have to illustrate
the use of OBI. Once calls are restarted, I think that should be the
primary focus. The approach taken of just doing the 'tissue sample from
organism' (or something like that) first will be fine.

One solution is documented here
http://docs.google.com/Doc?id=dhgmt5k6_2025vhhj4c3&hl=en_GB

Frank



 

- Bjoern

---

Mélanie Courtot

TFL- BCCRC

675 West 10th Avenue

Vancouver, BC

V5Z 1L3, Canada

[Frank Gibson]

On Mon, May 25, 2009 at 7:33 PM, Ryan Brinkman <rbri...@bccrc.ca> wrote:

I don’t disagree with you. I was wondering if there might be a way to avoid confusion by biologists who might confuse EOO/specimen with sample as the use the words interchangeably, perhaps by using another label for “sample”.

I would agree confusion can arise from these words that is why we have to be completely explicit and follow defintions in use.  Within the document I describe an EOO as any material entity the is_derived_from some organism. An EOO can be a sample, not all EOO are samples.  I hope this is intuitive.

Frank