[Fwd: [Fwd: Re: Environment Ontology]]

[Pankaj Jaiswal]

Hi Folks,
There is some more from Robin.

-Pankaj



-------- Original Message --------
   From: Robin Winter <rwinter at ich.ucl.ac.uk>
Subject: Re: Environment Ontology
     To: Pankaj Jaiswal <pj37 at cornell.edu>
     CC: pj37 at cornell.edu, rd120 at gen.cam.ac.uk

Dear Pankaj,

Looking at your scheme
a) The term 'genetic environment' is a bit confusing (unless you are trying to
bring in maternal effects etc.). Why not just 'genotype'

I think the two examples you give just reflect how the phenotype descriptors
(built into phenotypes) can be used.
Thus


> Pankaj - 1-In most of your cases that a physician records one may not know the
> genetic
> component
> So it will follow the path of A-B-F

Robin - This is the situation where a clinician is asking 'I wonder if phenotype
A in environment X has a genetic component ? ' (ie: it is a hypothetical
phenotype - the clinician needs to prove there is a genetic component)



> Pankaj - 2-In cases, where the genetic component is known then the path would
> depend on
> how a person proceeds. A-(C+B)-E or D-E.

Robin - In this case the clinicician will be saying 'the effect of genotype Q in
environment X is phenoype A'

We are just giving people the controlled vocabularies (phenotype
descriptor/environment/genotype) to build these statements
Robin


-------- Original Message --------


Hi Everyone,

For the last couple of days Robin, Pete, Rachel  and myself were discussing some
of the phenotype ontology details.
Please see the attached file phenotype_robin.txt, where Robin has tried to
outline some of his thoughts on phenotype ontology.


The text from our discussions is as follows....

Cheers

-Pankaj

-------- Original Message --------
   From: Robin Winter <rwinter at ich.ucl.ac.uk>
Subject: Re: Environment Ontology
     To: Pankaj Jaiswal <pj37 at cornell.edu>

Thanks for this,
I am glad I have started some debate !
Give me a bit of time to digest this and I will get back to you.
Robin


At 13:17 16/12/02 -0500, you wrote:

> Thanks Robin and Pete,
>
> If you remember, I presented the same upcoming strategy from Gramene, that a
> phenotype description is a collection of descriptors with its own attributes
> and
> values, instead of a one single phenotype term.
> Here is how we started curating the rice phenotypes associated to various
> genes
> and their alleles
> http://ascus.plbr.cornell.edu/~gramene/phenotype/Gramene-present.png
>
> But as we progressed, the strategies got modified based on the complexity and
> type of data that we were curating. This resulted in a modified approach,
> which
> we presented at the TIGR meeting.
> http://ascus.plbr.cornell.edu/~gramene/phenotype/Gramene-new.png
> In this scheme you will see that there are 2 types of phenotypes, simple and
> complex, which is the same as you have in your databases. Simple ones with
> only
> one descriptor where as the complex ones with one to many descriptors and a
> possible one-many gene interaction in a  given set of genetic and physical
> environments.
>
> Based on these modifications we can certainly create a multi dimensional or
> dynamic matrix comprising of
> -type of interaction among
> -genetic to genetic components AND/OR
> - genetic to environment components
> -the observable features
> -Anatomy
> -growth stage/ontogeny/developmental time line
> -Biochemical profiles (mol function/process/cellular component)
>
> Both of them will have their own attributes, values and scores, which
> determines
> the net observations for a phenotype.
>
> In that way you are right, we don't need a phenotype ontology. In order to
> build
> the various dynamic phenotype relationships every, database should have an
> anatomy to describe the organism on which it presents the data, the growth
> stages, various physical (biotic/abiotic) environments under which the
> responses/behavior/diseases for instance are recorded. The biochemical
> components will come from Gene Ontology. After identifying these components
> the
> question is how do we assess them, i.e. the attributes associated with them.
>
> e.g. for an enzyme, it could be its
> enzymatic activity/ Km
> inhibition/ Ki
> activation/ Ka
> structure (molecules structural integrity)
> sequence (modified sequence)
> etc.
>
> These attributes will have values of high/low/more/less on a comparative scale
>
> or on absolute scale of mM (milli molar) numerical values for Km etc. Similar
> to
> one records in biochemical analyses of Blood serum, urine samples. If I am not
>
> wrong, what we want is a uniform vocabulary of attributes and values that we
> want to associate to an observable feature.
>
> such as
>
> function and appearance
>
> Given this background here is the strategy, that I can think of
>
> http://ascus.plbr.cornell.edu/~gramene/phenotype/phenotype.png
>
> 1-In most of your cases that a physician records one may not know the genetic
> component
> So it will follow the path of A-B-F
>
> 2-In cases, where the genetic component is known then the path would depend on
>
> how a person proceeds. A-(C+B)-E or D-E.
>
> I think the situation #1 will solve the problem you suggested in meeting,
> about
> phalanges etc. Whereas in cases of syndrome you suggested, the situation #2
> should work.
>
> In plants #2 is certainly the case. #1 will appear only in cases where a
> randomly mutagenized population of individuals is present.
>
> By going through your notes it seems that you need to put up a controlled
> vocabulary of diseases and syndromes. The vocabulary already exists, all you
> may
> need is to put it in a DAG
> (http://www.geneontology.org/doc/GO.usage.html#General) to show multiple
> relationships for a complex disease.
>
> Best regards
> Pankaj
>
> Robin Winter wrote:
> >
> > Thanks,
> > I am not sure if people got the attached musings on syndromes/diseases etc.
> > Robin
> >
> > At 17:03 13/12/02 -0500, you wrote:
> >
> > > Hello Everyone,
> > >
> > > As we discussed at last weekend's phenotype ontology meeting at TIGR,
> > > Gramene
> > > Environment Ontology (GEO) is now available via GOBO site at
> > > http://www.geneontology.org/doc/gobo.html
> > > Pl. see the section on Environment.
> > >
> > > The GEO is in the developing stage, thus it needs your feedback.
> > >
> > > Cheers
> > > Pankaj
> > >
> > > ******************************************
> > > Pankaj Jaiswal, Ph.D.
> > > Dept. of Plant Breeding
> > > Cornell University
> > > Ithaca, NY-14853, USA
> > >
> > > Tel:+1-607-255-3103 / Fax:+1-607-255-6683
> > > E mail: pj37 at cornell.edu
> > > http://www.gramene.org
> > > ******************************************
> >
> > Prof. RM Winter
> > Dept Clinical and Molecular Genetics
> > Institute of Child Health
> > 30 Guilford Street
> > London WC1N  1EH
> > Tel: 0207 242 9789 ext. 2108
> > Fax: 0207 813 8141
> > Email: Rwinter at ich.ucl.ac.uk
> >
> >                      Name: Phenotypes.doc
> >    Phenotypes.doc    Type: Microsoft Word Document (application/msword)
> >                  Encoding: base64
>
> --
>
>

Prof. RM Winter
Dept Clinical and Molecular Genetics
Institute of Child Health
30 Guilford Street
London WC1N  1EH
Tel: 0207 242 9789 ext. 2108
Fax: 0207 813 8141
Email: Rwinter at ich.ucl.ac.uk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://brie4.cshl.edu/pipermail/gramene/attachments/20021217/865565aa/attachment.html>
-------------- next part --------------

!From Robin White. 
!Email: Rwinter at ich.ucl.ac.uk
!Copied from his original file sent to Pankaj on Saturday, 12/13/2002
____________________________________________________


Some thoughts on Phenotypes/Diseases/Syndromes etc.   

I was thinking about compound/complex phenotypes and don’t think we need these terms.

Presumably we would agree that the definition of a phenotype should be something like “the effect on an organism of an allele, or group of alleles, given a specific genetic background, and a specific environment”.

However, what we were actually trying to construct controlled vocabularies for are phenotype descriptors.  
These are individual units that can be put together to make up a phenotype.  
A phenotype is made up of one or more phenotype descriptor.
Different groups call phenotype descriptors different things (traits, symptoms, signs etc.). Confusion arises because some individual phenotype descriptors can be phenotypes in themselves.  
Thus, a cataract can be a phenotype descriptor, but can also be used on its own to define a phenotype.
The effect of some alleles on specific genetic backgrounds in mammals can be an isolated cataract (ie: cataract on its own with no other problems).  In this case the phenotype is ‘isolated cataract’ and it is made up of the single phenotype descriptor ‘cataract’
On the other hand there may be a syndrome (eg: Winter Syndrome) which includes cataracts and deafness which is caused a mutant allele. In this case the phenotype is ‘Winter syndrome’ and this is defined by the phenotype descriptors ‘cataract’ and ‘deafness’

If we accept that phenotypes are made up of one or more phenotype descriptor, then we don’t need controlled vocabularies for phenotypes (or diseases or syndromes), we just need controlled vocabularies for phenotype descriptors.  
Note that some diseases are phenotypes (i.e. collections of phenotype descriptors), and some not).  Marfan syndrome is a disease and a phenotype, because it is the effect of a single allele causing the combination of descriptors long fingers, dislocation of the lens, heart valve abnormalities etc.

Malaria is not a phenotype, because it is not the consequence of an allele or alleles.  On the other hand, susceptibility to malaria is a phenotype (as described by the phenotype descriptors for various antigen responses).

The consequence of all this is that we do not need the concept of complex or compound phenotypes (a phenotype is defined as a combination of one or more phenotype descriptors).  We also do not need a controlled vocabulary for diseases or syndromes, as if these are phenotypes, they are just made up of combinations of phenotype descriptors from a controlled vocabulary.  

It also means that the controlled vocabulary for phenotypes does not need to include terms to describe the environment.  Terms to describe the environment need their own controlled vocabulary.  The phenotype descriptors will however include terms for response to environment.  In the example of ‘response to malaria’, the environmental descriptor might be ‘antigen challenge’ whereas the phenotype descriptor is ‘response to antigen challenge’.

Does this make sense ?