Quantcast

[Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

classic Classic list List threaded Threaded
15 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

[Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
I have a few questions concerning these tables.  I would like to store a protocol of 'NaCl treatment', reagent 'NaCl', and a property of 'concentration' with value '100' and unit 'micromolar'.

First, regarding nd_protocol, can we remove the UNIQUE NOT NULL constraint on name?  The NaCl treatment is applied in every experiment so the name is the same.  I think also there should be a type_id column because that's what I would use to store the name of the protocol.

I'm planning to put the concentration property in nd_protocolprop.  I propose to add a units_id column because this is a quantitative property and I can imagine this is a common use case for protocols.  In addition I propose to add a units_id column to projectprop and nd_experimentprop because I have quantitative properties in projectprop and I can forsee someone having them for nd_experimentprop as well.

For nd_protocol_reagent, how is the type_id used?  In this case what would be a good term for it?

I'd also like to propose adding environmentprop and phenstatementprop tables to make statements like 'NaCL treatment of 100 micromolar (E) for genotype G, phenotype P produces  average value of X'. 


thanks,

yuri

------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Emmanuel G. Dialynas
Hi Yuri,

This is how we are storing reagents concentrations for insecticide
resistance experiments in VectorBase.
We have two terms in VBCV (VectorBase Controlled Vocabulary),
'Concentration unit' and 'Concentration value'.
The value of the 'Concentration unit' cvterm is an ontology term from
the Units of Measurement Ontology (UO)
and the value for the 'Concentration value' cvterm is the actual value
of the concentration.
The above data are stored in nd_reagentprop.

Yes, removing the UNIQUE_NOT_NULL constraint on name would be useful for
me too.

Cheers,
Emmanuel



On 12/16/2010 01:32 AM, Yuri Bendana wrote:

> I have a few questions concerning these tables.  I would like to store
> a protocol of 'NaCl treatment', reagent 'NaCl', and a property of
> 'concentration' with value '100' and unit 'micromolar'.
>
> First, regarding nd_protocol, can we remove the UNIQUE NOT NULL
> constraint on name?  The NaCl treatment is applied in every experiment
> so the name is the same.  I think also there should be a type_id
> column because that's what I would use to store the name of the protocol.
>
> I'm planning to put the concentration property in nd_protocolprop.  I
> propose to add a units_id column because this is a quantitative
> property and I can imagine this is a common use case for protocols.
>  In addition I propose to add a units_id column to projectprop and
> nd_experimentprop because I have quantitative properties in
> projectprop and I can forsee someone having them for nd_experimentprop
> as well.
>
> For nd_protocol_reagent, how is the type_id used?  In this case what
> would be a good term for it?
>
> I'd also like to propose adding environmentprop and phenstatementprop
> tables to make statements like 'NaCL treatment of 100 micromolar (E)
> for genotype G, phenotype P produces  average value of X'.
>
>
> thanks,
>
> yuri
>
>
> ------------------------------------------------------------------------------
> Lotusphere 2011
> Register now for Lotusphere 2011 and learn how
> to connect the dots, take your collaborative environment
> to the next level, and enter the era of Social Business.
> http://p.sf.net/sfu/lotusphere-d2d
>
>
> _______________________________________________
> Gmod-phendiver mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/gmod-phendiver


------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

seth redmond
If the concentration is the only thing that changes between assays,  
then surely this should be stored in nd_experimentprop? It is an  
experimental factor after all. You'd then have a single entry in  
protocol for "NaCL treatment", a varying concentration in the  
nd_experimentprop, no redundancy, and no requirement for a non-unique  
name.

As to the units field. I'll go with the majority, but I feel  
Emmanuel's approach is the right one here. Whilst there might be a use  
for units in some of the prop fields, it's not clear (as it was in the  
case of the phenotype module) that these will be used by a sufficient  
number of people for it to be worth making changes to so many tables  
in the schema. I'd much rather leave the prop tables lightweight and  
flexible.

I'm also struggling to think of an example for numerical data in  
projectprop; Yuri, could you give an example?

-s



On 16 Dec 2010, at 11:40, Emmanuel G. Dialynas wrote:

> Hi Yuri,
>
> This is how we are storing reagents concentrations for insecticide
> resistance experiments in VectorBase.
> We have two terms in VBCV (VectorBase Controlled Vocabulary),
> 'Concentration unit' and 'Concentration value'.
> The value of the 'Concentration unit' cvterm is an ontology term from
> the Units of Measurement Ontology (UO)
> and the value for the 'Concentration value' cvterm is the actual value
> of the concentration.
> The above data are stored in nd_reagentprop.
>
> Yes, removing the UNIQUE_NOT_NULL constraint on name would be useful  
> for
> me too.
>
> Cheers,
> Emmanuel
>
>
>
> On 12/16/2010 01:32 AM, Yuri Bendana wrote:
>> I have a few questions concerning these tables.  I would like to  
>> store
>> a protocol of 'NaCl treatment', reagent 'NaCl', and a property of
>> 'concentration' with value '100' and unit 'micromolar'.
>>
>> First, regarding nd_protocol, can we remove the UNIQUE NOT NULL
>> constraint on name?  The NaCl treatment is applied in every  
>> experiment
>> so the name is the same.  I think also there should be a type_id
>> column because that's what I would use to store the name of the  
>> protocol.
>>
>> I'm planning to put the concentration property in nd_protocolprop.  I
>> propose to add a units_id column because this is a quantitative
>> property and I can imagine this is a common use case for protocols.
>> In addition I propose to add a units_id column to projectprop and
>> nd_experimentprop because I have quantitative properties in
>> projectprop and I can forsee someone having them for  
>> nd_experimentprop
>> as well.
>>
>> For nd_protocol_reagent, how is the type_id used?  In this case what
>> would be a good term for it?
>>
>> I'd also like to propose adding environmentprop and phenstatementprop
>> tables to make statements like 'NaCL treatment of 100 micromolar (E)
>> for genotype G, phenotype P produces  average value of X'.
>>
>>
>> thanks,
>>
>> yuri
>>
>>
>> ------------------------------------------------------------------------------
>> Lotusphere 2011
>> Register now for Lotusphere 2011 and learn how
>> to connect the dots, take your collaborative environment
>> to the next level, and enter the era of Social Business.
>> http://p.sf.net/sfu/lotusphere-d2d
>>
>>
>> _______________________________________________
>> Gmod-phendiver mailing list
>> [hidden email]
>> https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
>
>
> ------------------------------------------------------------------------------
> Lotusphere 2011
> Register now for Lotusphere 2011 and learn how
> to connect the dots, take your collaborative environment
> to the next level, and enter the era of Social Business.
> http://p.sf.net/sfu/lotusphere-d2d
> _______________________________________________
> Gmod-schema mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/gmod-schema


------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Naama Menda
In reply to this post by Yuri Bendana-3
hi Yuri,
answering inline:

On Wed, Dec 15, 2010 at 6:32 PM, Yuri Bendana <[hidden email]> wrote:
I have a few questions concerning these tables.  I would like to store a protocol of 'NaCl treatment', reagent 'NaCl', and a property of 'concentration' with value '100' and unit 'micromolar'.

First, regarding nd_protocol, can we remove the UNIQUE NOT NULL constraint on name?  The NaCl treatment is applied in every experiment so the name is the same.  I think also there should be a type_id column because that's what I would use to store the name of the protocol.

I agree with Seth. This is the same protocol, with different nd_experimentprop

 
I'm planning to put the concentration property in nd_protocolprop.  I propose to add a units_id column because this is a quantitative property and I can imagine this is a common use case for protocols.  In addition I propose to add a units_id column to projectprop and nd_experimentprop because I have quantitative properties in projectprop and I can forsee someone having them for nd_experimentprop as well.

Would using , like Emmanuel does, UO terms for the unit work for you? Your use case is very specific for this type of experiment, and the prop tables should be kept very generic
 
For nd_protocol_reagent, how is the type_id used?  In this case what would be a good term for it?

I'd also like to propose adding environmentprop and phenstatementprop tables to make statements like 'NaCL treatment of 100 micromolar (E) for genotype G, phenotype P produces  average value of X'. 


In this case, what would you store in the proposed prop tables? Does 'value X' refer to the value of 'phenotype P' ? or are you talking about pleiotropic phenotypes ?

 
thanks,

yuri

-Naama
 

------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
Hi Naama, Seth, and Emmanuel,

Thanks for the responses.

I was thinking nd_protocol would be like phenotype where any values are stored in the associated prop table and that   nd_experimentprop is for experimental properties that don't have an associated parent table.  But this is fine, I like the no redundancy.  I still would like a type_id in nd_protocol in addition to the name since I define the protocol in my ontology.

Emmanuel:  I don't quite follow how you store multiple values of 'Concentration value' and 'Concentration unit' in nd_reagentprop.  Do you have more than one protocol in nd_protocol?  Is it really the same protocol with just unique names?

I still think adding units_id is a good idea to any prop tables that could hold quantitative data (by the way, I am using the UO for the units_id).  Otherwise you have the situation described by Emmanuel where you need to input two rows, one for the value and the other for the unit (and have two cvterms respectively).  That's pretty awkward.  Adding a nullable integer column for units is lightweight and generic in my view.  I'm not saying we have to add units_id to all the prop tables in the schema, just the ones where it could be convenient.  Right now I see that as projectprop and nd_experimentprop (but see below).  

Seth, an example of a quantitative property I have in projectprop is: "Watering Amount: 100 ml".

By the way, just curious, what is the feature_id in nd_reagent for?



I'd also like to propose adding environmentprop and phenstatementprop tables to make statements like 'NaCL treatment of 100 micromolar (E) for genotype G, phenotype P produces  average value of X'. 


In this case, what would you store in the proposed prop tables? Does 'value X' refer to the value of 'phenotype P' ? or are you talking about pleiotropic phenotypes ?

 In the environmentprop I would store the property 'NaCl treatment' with value of 100 micromolar (this table would be another good candidate for units_id).  In phenstatement I would set the type_id to 'summary statistic'. In phenstatementprop I would store the property 'mean' with value of X  (units_id here too). This would be the average value of the phenotype P given the environment E and genotype G.

yuri


------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
Actually, now that I think about it, this situation with nd_protocol is similar to the one I had with phenotype.  Yes, 'concentration' is a property of the experiment but it also is a property of the protocol.  So, like with phenotype when I created a nd_experiment_phenotypeprop to store the phenotype values of an experiment, my vote here is to create a nd_experiment_protocolprop.  This avoids the data redundancy of storing the values in nd_protocolprop but also is less ambiguous than storing it in nd_experimentprop (think of the situation of having multiple protocols with multiple concentrations and having to keep track of which is which using the rank of the 'concentration' property or making custom concentration cvterms).  I think nd_experiment_protocolprop is a cleaner design.

yuri

On Thu, Dec 16, 2010 at 7:42 PM, Yuri Bendana <[hidden email]> wrote:
Hi Naama, Seth, and Emmanuel,

Thanks for the responses.

I was thinking nd_protocol would be like phenotype where any values are stored in the associated prop table and that   nd_experimentprop is for experimental properties that don't have an associated parent table.  But this is fine, I like the no redundancy.  I still would like a type_id in nd_protocol in addition to the name since I define the protocol in my ontology.

Emmanuel:  I don't quite follow how you store multiple values of 'Concentration value' and 'Concentration unit' in nd_reagentprop.  Do you have more than one protocol in nd_protocol?  Is it really the same protocol with just unique names?

I still think adding units_id is a good idea to any prop tables that could hold quantitative data (by the way, I am using the UO for the units_id).  Otherwise you have the situation described by Emmanuel where you need to input two rows, one for the value and the other for the unit (and have two cvterms respectively).  That's pretty awkward.  Adding a nullable integer column for units is lightweight and generic in my view.  I'm not saying we have to add units_id to all the prop tables in the schema, just the ones where it could be convenient.  Right now I see that as projectprop and nd_experimentprop (but see below).  

Seth, an example of a quantitative property I have in projectprop is: "Watering Amount: 100 ml".

By the way, just curious, what is the feature_id in nd_reagent for?



I'd also like to propose adding environmentprop and phenstatementprop tables to make statements like 'NaCL treatment of 100 micromolar (E) for genotype G, phenotype P produces  average value of X'. 


In this case, what would you store in the proposed prop tables? Does 'value X' refer to the value of 'phenotype P' ? or are you talking about pleiotropic phenotypes ?

 In the environmentprop I would store the property 'NaCl treatment' with value of 100 micromolar (this table would be another good candidate for units_id).  In phenstatement I would set the type_id to 'summary statistic'. In phenstatementprop I would store the property 'mean' with value of X  (units_id here too). This would be the average value of the phenotype P given the environment E and genotype G.

yuri



------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Sook Jung
In reply to this post by Yuri Bendana-3
 
Hi Yuri,
Answers and comments below.
 
By the way, just curious, what is the feature_id in nd_reagent for?
Those tables like nd_experiment, nd_protocol and nd_reagents were adopted from the initial version of nd module, developed for Helicounious DB, without very much further discussion.
The initial version with detailed use of the tables are available below.
 
From what I remember, those tables were mostly for genotyping experiments, where the protocol table was for storing markers and the reagent table was storing primers, adaptors, etc. So the feature_id in reagent table was to link to the primer sequences stored in the feature table. There was also feature_protocol table to link the marker sequence in feature table (locus). The use of these tables this way is up to the users I think. We were going to use these tables this way, but now we'll store markers in feature table. The same marker can be used with different protocol (different PCR protocol, etc) in each experiment, so the protocol table could be used to store those type of data.
 
I have a question. Is the NaCl treatment is part of sample treatment or part of the phenotyping assay? In other words, do you have samples treated with the same NaCl conc that will be further tested for multiple phenotyping assays? We store all the samples in stock table - from population down to samples - so the sample treatment property goes to stockprop table. The detailed definition of the sample treatment could be stored in cvterm. The nd_experiment is either to store phenotyping or genotyping (or field collection) experiment and the protocol would be the detailed protocol used for those phenotyping or genotyping experiment.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.
 
Thanks
 
Sook
 

------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Naama Menda
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama



------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama




------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

seth redmond
On the question of where to store reagent_concentrations; I don't think where we store reagent_conc values is actually the issue. It's a question of *how* we store them. For instance if we have a single experimental factor then storing it in experimentprop or experiment_protocolprop is equally unambiguous. On the other hand if there are two experimental factors for a single protocol then the experiment_protocolprop won't actually make this any clearer (we actually have a number of cases like this: insecticide resistance assays where we have an 'insecticide' and a 'synergist' both of which are variable). 

There are two ways around this that I can see. The first is to create an experiment_reagent and experiment_reagentprop table, which could clearly identify which reagent conc was being changed. This seems like a neat solution, but I'm actually dead against it for two reasons: 
1) it will be much less clear to the novice user which connection they should make - I'd much prefer to force people to make a protocol before they can start linking reagents
2) if I find out the experimental factor is NaCl conc, I'd much rather *know* that it's going to be in experimentprop than have to go hunting for it in the reagentprop / protocolprop / experiment_protocolprop / etc, tables. 

The other solution is to define your experimental factors more clearly - either by using specific terms in an ontology: [insecticide_concentration / synergist concentration] or for two insecticides [DDT concentration / DEF concentration] 
Alternatively if users do not want to develop sufficiently complex ontologies we could achieve the same effect by using the rank column in the various tables.

 store in nd reagentprop:
CVTERM VALUE RANK
DDT (insecticide) \N 1
concentration units percent 1
DEF (insecticide) \N 2
concentration units percent 2

and in nd_experiment_prop
CVTERM VALUE RANK
concentration 1 1
concentration 0.01 2

-s



On 17 Dec 2010, at 21:25, Yuri Bendana wrote:

Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama



<ATT00003..txt><ATT00004..txt>


------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Naama Menda
In reply to this post by Yuri Bendana-3
the schema should work for both use cases. I don't think there is one solid definition for nd_experiment, and based on your data you should be able to store one experiment per assay or one experiment per sample , and so on.

As for NaCl concentration, if you want to store it as a protocol, then yes, nd_experiment_protocolprop is a good place for the values.
Out of curiosity, what other protocols are you storing? (or anyone else who's using the nd_protocol table)

-Naama

 
On Fri, Dec 17, 2010 at 4:25 PM, Yuri Bendana <[hidden email]> wrote:
Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama





------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
In reply to this post by seth redmond
I don't agree completely with this statement because if you store a concentration in nd_experiment_protocolprop you have linked it to that particular protocol although not to a specific reagent as you mention (is 'experimental factor' the same as a reagent?).  As you described you could use rank to distinguish between the concentration of multiple reagents in a protocol.

But if you just store the concentration in nd_experimentprop, you have less information about which protocol it refers to, much less the reagent.  It might not even refer to a protocol.  To me it's more ambiguous.

yuri

On Mon, Dec 20, 2010 at 9:42 AM, seth redmond <[hidden email]> wrote:
On the question of where to store reagent_concentrations; I don't think where we store reagent_conc values is actually the issue. It's a question of *how* we store them. For instance if we have a single experimental factor then storing it in experimentprop or experiment_protocolprop is equally unambiguous. On the other hand if there are two experimental factors for a single protocol then the experiment_protocolprop won't actually make this any clearer (we actually have a number of cases like this: insecticide resistance assays where we have an 'insecticide' and a 'synergist' both of which are variable). 



On 17 Dec 2010, at 21:25, Yuri Bendana wrote:

Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama



<ATT00003..txt><ATT00004..txt>



------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
In reply to this post by Naama Menda
Yes, the schema does work for both use cases now, which is great.  I was just trying to clear the confusion about which use case we're implementing.  Otherwise the changes I'm proposing won't make sense.

For this nd project I'm dealing with right now, there is only one protocol.  The others may have more than one, like a Rhizobia treatment.

yuri

On Mon, Dec 20, 2010 at 11:40 AM, Naama Menda <[hidden email]> wrote:
the schema should work for both use cases. I don't think there is one solid definition for nd_experiment, and based on your data you should be able to store one experiment per assay or one experiment per sample , and so on.

As for NaCl concentration, if you want to store it as a protocol, then yes, nd_experiment_protocolprop is a good place for the values.
Out of curiosity, what other protocols are you storing? (or anyone else who's using the nd_protocol table)

-Naama


 
On Fri, Dec 17, 2010 at 4:25 PM, Yuri Bendana <[hidden email]> wrote:
Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama






------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

seth redmond
In reply to this post by Yuri Bendana-3
Well, as I think I've said before, we're using nd_experiments as single assays: one assay, one protocol. Used in this way there's no ambiguity over what protocol is being referred to. 
Incidentally, whilst I've no great objections to other people using multiple protocols for a single nd_experiment, it's worth bearing in mind that if you subsequently link the experiment to a pheno/genotype it's also not obvious which protocol generated that phenotype - you're relying on the perfect knowledge of your users to make that call. Personally I don't think this is a great idea, but YMMV. 
As a bonus, of course, this obviates the need for yet another prop table, and ensures that all variables relevant that assay are stored in the same place.

Basically I think in both cases - multiple protocols to one experiment, or multiple reagents to one protocol - we have two choices:
1) add tables to the schema to make it unambiguous (nd_experiment_protocolprop, nd_experiment_reagent and nd_experiment_reagentprop should do it. Oh. and if you need to specify one-of-multiple protocols for a phenotype/genotype, you'll be needing an nd_experiment_phenotypeprop and nd_experiment_genotypeprop too)
2) use the ontologies and rank columns to solve ambiguities, keep the schema lightweight, and expect those who have multiple protocols or reagents to develop the ontologies to handle it. 


On 20 Dec 2010, at 20:56, Yuri Bendana wrote:

I don't agree completely with this statement because if you store a concentration in nd_experiment_protocolprop you have linked it to that particular protocol although not to a specific reagent as you mention (is 'experimental factor' the same as a reagent?).  As you described you could use rank to distinguish between the concentration of multiple reagents in a protocol.

But if you just store the concentration in nd_experimentprop, you have less information about which protocol it refers to, much less the reagent.  It might not even refer to a protocol.  To me it's more ambiguous.

yuri

On Mon, Dec 20, 2010 at 9:42 AM, seth redmond <[hidden email]> wrote:
On the question of where to store reagent_concentrations; I don't think where we store reagent_conc values is actually the issue. It's a question of *how* we store them. For instance if we have a single experimental factor then storing it in experimentprop or experiment_protocolprop is equally unambiguous. On the other hand if there are two experimental factors for a single protocol then the experiment_protocolprop won't actually make this any clearer (we actually have a number of cases like this: insecticide resistance assays where we have an 'insecticide' and a 'synergist' both of which are variable). 



On 17 Dec 2010, at 21:25, Yuri Bendana wrote:

Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama



<ATT00003..txt><ATT00004..txt>




------------------------------------------------------------------------------
Lotusphere 2011
Register now for Lotusphere 2011 and learn how
to connect the dots, take your collaborative environment
to the next level, and enter the era of Social Business.
http://p.sf.net/sfu/lotusphere-d2d
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [Gmod-phendiver] [Gmod-schema] nd_protocol, nd_reagent, nd_protocol_reagent, environment, phenstatement

Yuri Bendana-3
I forgot you're using the assay-experiment, now it's clearer what you mean.

I would go with adding the prop tables.  They are small and lightweight, basically just extensions of the base tables and in a relatively uniform format.  I suggest any new ones plus projectprop and nd_experimentprop should have units_id columns.

The implementer can decide which combination of prop tables and ontologies/rank to use.  This makes the schema very flexible and extensible, which I think has higher priority than making it light as possible since we all have different use cases.

In my case, I'll use nd_experiment_protocolprop and rank.  I don't anticipate using nd_experiment_reagent and nd_experiment_reagentprop.  Like you said, I'd rather link to the reagents thru their protocols rather than directly with experiment.   Right now, I don't have protocols recorded for every phenotype observation, just for the treatments.

nd_experiment_phenotypeprop would come in handy for storing phenotype values and linking to specific protocols in my case. I can see nd_experiment_genotypeprop also being useful if I were storing genotyping experiments.

yuri

On Tue, Dec 21, 2010 at 12:04 AM, seth redmond <[hidden email]> wrote:
Well, as I think I've said before, we're using nd_experiments as single assays: one assay, one protocol. Used in this way there's no ambiguity over what protocol is being referred to. 
Incidentally, whilst I've no great objections to other people using multiple protocols for a single nd_experiment, it's worth bearing in mind that if you subsequently link the experiment to a pheno/genotype it's also not obvious which protocol generated that phenotype - you're relying on the perfect knowledge of your users to make that call. Personally I don't think this is a great idea, but YMMV. 
As a bonus, of course, this obviates the need for yet another prop table, and ensures that all variables relevant that assay are stored in the same place.

Basically I think in both cases - multiple protocols to one experiment, or multiple reagents to one protocol - we have two choices:
1) add tables to the schema to make it unambiguous (nd_experiment_protocolprop, nd_experiment_reagent and nd_experiment_reagentprop should do it. Oh. and if you need to specify one-of-multiple protocols for a phenotype/genotype, you'll be needing an nd_experiment_phenotypeprop and nd_experiment_genotypeprop too)
2) use the ontologies and rank columns to solve ambiguities, keep the schema lightweight, and expect those who have multiple protocols or reagents to develop the ontologies to handle it. 


On 20 Dec 2010, at 20:56, Yuri Bendana wrote:

I don't agree completely with this statement because if you store a concentration in nd_experiment_protocolprop you have linked it to that particular protocol although not to a specific reagent as you mention (is 'experimental factor' the same as a reagent?).  As you described you could use rank to distinguish between the concentration of multiple reagents in a protocol.

But if you just store the concentration in nd_experimentprop, you have less information about which protocol it refers to, much less the reagent.  It might not even refer to a protocol.  To me it's more ambiguous.

yuri

On Mon, Dec 20, 2010 at 9:42 AM, seth redmond <[hidden email]> wrote:
On the question of where to store reagent_concentrations; I don't think where we store reagent_conc values is actually the issue. It's a question of *how* we store them. For instance if we have a single experimental factor then storing it in experimentprop or experiment_protocolprop is equally unambiguous. On the other hand if there are two experimental factors for a single protocol then the experiment_protocolprop won't actually make this any clearer (we actually have a number of cases like this: insecticide resistance assays where we have an 'insecticide' and a 'synergist' both of which are variable). 



On 17 Dec 2010, at 21:25, Yuri Bendana wrote:

Hi Sook and Naama,

It sounds like, unfortunately, we are still not on common ground on the definition of an 'experiment'.  For us, an experiment is the collection of observations and treatments done on one individual (or sample).  So, an experiment is sample-centric rather than assay-centric.

Given this definition of experiment, I originally put the value of 'NaCl concentration' in the nd_experiment_stockprop.  But Maren told me that 'NaCl treatment' is actually a protocol, so that's why I'm trying to put the treatment definition there.  But the concentration values I think belong in a nd_experiment_protocolprop rather than in nd_protocol as I was proposing before and nd_experimentprop as has been suggested.

Thanks everyone for your responses.

yuri

On Fri, Dec 17, 2010 at 8:11 AM, Naama Menda <[hidden email]> wrote:
On Fri, Dec 17, 2010 at 2:17 AM, Sook Jung <[hidden email]> wrote:
 
Hi Yuri,
Answers and comments below.
 
I remember, though, we decided to allow linking one nd_experiment to mutilple phenotype rows. In that case, one nd_experiment_id really represent a sample, not a single experiment. Then you could store samples (that are treated with a certain NaCl conc)  in nd_experiment and store the NaCl conc. in nd_experimentprop. If the sample treatment is complex (more than NaCl conc), the NaCl conc could be stored in nd_protocol.. I think it will help if we have more detail on your data.

yes, if this is one 'experiment' (NaCl treatment) then the concentration is the experimentprop (or experiment_protocolprop ) and all subsequent phenptypes/genotypes should be linked to the same nd_experiment_id .

I agree with Sook, it would help to see a full data set of your experiments

-Naama



<ATT00003..txt><ATT00004..txt>





------------------------------------------------------------------------------
Forrester recently released a report on the Return on Investment (ROI) of
Google Apps. They found a 300% ROI, 38%-56% cost savings, and break-even
within 7 months.  Over 3 million businesses have gone Google with Google Apps:
an online email calendar, and document program that's accessible from your
browser. Read the Forrester report: http://p.sf.net/sfu/googleapps-sfnew
_______________________________________________
Gmod-phendiver mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/gmod-phendiver
Loading...