Re: [Obo-relations] FuGE, Ontologies, NINDS Microarray, GENSAT, ABA, and AD

Hi Michael,

Sorry, I'm a little bit behind reading/responding emails these days. 
Please see my response below.

Miller, Michael D (Rosetta) wrote:
> hi kai,
>
>   
>> I think these are related subjects
>>     
>
> well, as we've seen, everything is related to everything else, but,
> specifically, there's obviously the need for people to be able to not
> only compare microarray experiments for similarities but all types of
> system biology experiments, one interesting use case (among many) is
> comparing proteomic results with gene expression results for the same
> set of subjects.
>   
Web science I think is about relating many things including communities 
and people. I gave a few slides to Dr. Mane in the hope that he would be 
able convince folks to think about SW at the NINDS Microarray Consortium 
meeting which will be held within a couple of days. It's great that you 
mentioned proteomic data. I used GENSAT only as an example, but we can 
extend to other types of data including proteomic data. In addition to 
NINDS microarray center, Yale has a neuroproteomic center funded by NIDA 
(National Institute on Drug Abuse). We are actually planning to 
correlate gene expression data with protein expression data.

>> Before going into further 
>> detail about how to convert NINDS microarray data into OWL ...
>>     
>
> one item i was unclear on was what the scope of conversion of microarray
> data into OWL you have in mind.  my take on this, as i said previously,
> is that in MAGE and eventually in MAGEv2, the actual performance of the
> microarray experiment and the results are captured in the XML schema
> generated from the UML model.  there is also the ability to annotate,
> using the ontology package, all the objects in the XML document.  it's
> this annotation i believe can be easily translated into RDF or OWL,
> based on the early version of the OMG Ontology Metamodel.  the rest of
> the document is easily comparable with other documents at the XML or UML
> level.
>   
At this point, we don't need to worry about the measurement data (which 
is enormous). We can start with experiment and sample annotation. "Are 
current ontologies in biology good ontologies?". I hope, as a community, 
we can prove that they are.

Cheers,

-Kei
> cheers,
> michael
>
>   
>> -----Original Message-----
>> From: Kei Cheung [mailto:kei.cheung@yale.edu] 
>> Sent: Saturday, June 02, 2007 7:26 AM
>> To: Miller, Michael D (Rosetta)
>> Cc: William Bug; Smith, Barry; Kashyap, Vipul; 
>> samwald@gmx.at; public-semweb-lifesci@w3.org; 
>> obo-relations@lists.sourceforge.net; Paul Spellman; Shrikant Mane
>> Subject: Re: [Obo-relations] FuGE, Ontologies, NINDS 
>> Microarray, GENSAT, ABA, and AD
>>
>> Hi Michael, Bill, Barry, et al.,
>>
>> I think these are related subjects (see the modified title and 
>> discussion below). I also copied this email to the Yale PI 
>> (Dr. Shrikant 
>> Mane) who is one of the PI's of the NINDS microarray 
>> consortium (Yale is 
>> among one of the four centers in the consortium) in the hope of 
>> initiating an interaction between the consortium and 
>> SWHCLS/ontology/MGED groups. I'm also trying to practice TBL's "Web 
>> Science" synergizing different communities. Before going into further 
>> detail about how to convert NINDS microarray data into OWL , 
>> we should 
>> probably step back and see what we might accomplish.
>>
>> Since our HCLS paper (http://www.biomedcentral.com/1471-2105/8/S3/S2) 
>> and Banff demo (http://esw.w3.org/topic/HCLS/Banff2007Demo) 
>> centered on 
>> a use case of Alzheimer Disease (AD), for the fun of it, I searched 
>> using the keyword "Alzheimer Disease" to see if there are any AD 
>> microarray data stored in the NINDS microarray consortium repository. 
>> It's not a surprise that I found a number of neuroscience microarray 
>> projects that have to do with the study of AD. Below is the 
>> URL pointing 
>> to the description of one of the AD microarray projects.
>>
>> http://arrayconsortium.tgen.org/np2/viewProject.do?action=view
>> Project&projectId=110
>>
>> As you can see the content is structured and I think can be converted 
>> into an OWL ontology. More interestingly, it seems to have some 
>> annotation errors. For example, if you look at the following 
>> field/value 
>> pairs:
>>
>> organ/tissue type: blood
>> organ region: adrenal medulla
>>
>> According to the description of the project (e.g., experimental 
>> procedure and design), I think they should be:
>>
>> organ/tissue type: brain
>> organ region: entorhinal cortex
>>
>> Basically, this microarray project generated differential gene 
>> expression profiling data for neurons containing 
>> neurofibrillary tangles 
>> and normal neurons from the entorhinal cortex of AD cases (mid-stage).
>>
>> Even more interestingly, one can integrate such gene expression data 
>> with other types of gene expression data (e.g., image-based gene 
>> expression data) stored in other repositiories. For example, 
>> if you go 
>> to GENSAT, you can find what genes are expressed in a given 
>> brain region 
>> (e.g., entorhinal cortex).
>>
>> http://braininfo.rprc.washington.edu/Scripts/indexothersite.as
>> px?ID=150&type=h&term=entorhinal_area&thterm=Entorhinal%20Cort
>>     
> ex&city=Bethesda&country=USA&institue=NCBI,%20National%20Library%20of%>
> 20Medicine&namesite=GENSAT&URL=http://www.ncbi.nlm.nih.gov/pro
>   
>> jects/gensat/index.cgi?cmd=searchhjsAMPgene_name=hjsAMPgene_sy
>>     
> m=hjsAMPage=anyhjsAMPexp_tech=anyhjsAMPplane=anyhjsAMPregion=Entorhinal+
> CortexhjsAMPcell=anyhjsAMPl> evel=anyhjsAMPfmt=gene
>   
>> Of course, similar integration can be performed with other 
>> image-based 
>> gene expression repositories such as Allen Brain Atlas.
>>
>> Cheers,
>>
>> -Kei
>>
>> Miller, Michael D (Rosetta) wrote:
>>
>>     
>>> hi bill and kei,
>>>  
>>> i've changed the subject, since this is moving away from 
>>>       
>> the original 
>>     
>>> topic.
>>>  
>>> "Yes - you are right, of course - right now the TGEN infrastructure 
>>> for the consortium is committed to providing MAGE-ML instances [1]."
>>>  
>>> that's great.
>>>  
>>> "the FuGE-stk [2] will provide a means to "convert" MAGE-ML 
>>>       
>> to FuGE-ML"
>>     
>>>  
>>> not exactly, the folks (myself included) that worked on 
>>>       
>> FuGE but with 
>>     
>>> a focus on microarrays are working on MAGEv2 and there is a 
>>>       
>> commitment 
>>     
>>> to provide a way to translate to/from MAGEv1 <-> MAGEv2.  at the 
>>> minimum this would be a mapping but if there is time and resources 
>>> available, this would also have an implementation in the 
>>>       
>> MAGEstk v2. 
>>     
>>>  
>>> MAGEv2 is being built on top of FuGE as an extension to add in 
>>> microarray specific classes (extending Data as ArrayDesign, 
>>> DesignElementData, etc, Material as Array, QPCRPlate, etc, and 
>>> DimensionElement as DesignElement extended by Feature, 
>>>       
>> Reporter, and 
>>     
>>> CompositeElement).
>>>  
>>> we have not been quite as organized as the MAGEv1 effort so 
>>>       
>> the work 
>>     
>>> doesn't have a high visibility yet, plus we have been working on 
>>> MAGE-TAB, a simplified, spreadsheet version of the MAGE-OM model.
>>>  
>>> i'm hoping we can get back on track soon, we are not that far from 
>>> completion, perhaps the NIH or BIRN microarray folks would 
>>>       
>> be willing 
>>     
>>> to host a MAGEv2 meeting? (note, this would have little to do with 
>>> ontology development!)
>>>  
>>> "many of the experts working on FuGE .. are looking for 
>>>       
>> assistance in 
>>     
>>> how to make use of ontologies when representing microarray 
>>>       
>> data in a 
>>     
>>> FuGE instance"
>>>  
>>> yes, but note that this has nothing to do with ontology 
>>>       
>> modeling per 
>>     
>>> se but simply the best way to model ontology annotation for the 
>>> objects of a FuGE document.  in essence, a FuGE object, such as a 
>>> Material that represents a rat or the Investigation itself, becomes 
>>> either implicitly or explicitly an Individual of the 
>>>       
>> desired classes 
>>     
>>> from whatever ontologies that are appropriate.  it then 
>>>       
>> inherits the 
>>     
>>> properties of those classes (if there are any) and can specify slot 
>>> instances.  it is anticipated that OBI will be usable for 
>>>       
>> most of the 
>>     
>>> basic annotation then, perhaps, specialized ontologies in 
>>>       
>> the domain 
>>     
>>> of the particular investigation can annotation more exactly.
>>>  
>>> most specifically, information and relationships of these 
>>> referenced classes would not be in the FuGE document, just the 
>>> information necessary to look them up in the ontology itself.
>>>  
>>> we modeled the FuGE ontology package from the Individual 
>>>       
>> diagram of an 
>>     
>>> early draft of OMG's Ontology Definition MetaModel (ODM).  that 
>>> section was actually explicitly dropped from the final 
>>>       
>> version of the 
>>     
>>> ODM because it had problems with OWL Full (and perhaps DL), but we 
>>> anticipate that the vast majority of desired ontology 
>>>       
>> annotation can 
>>     
>>> be captured via this model.
>>>  
>>> temporal and containment/association relationships are actually 
>>> intended to be captured by the FuGE objects (the flow of 
>>>       
>> Material and 
>>     
>>> Data through ProtocolApplications, the various associations between 
>>> FuGE classes)
>>>  
>>> "there is both an eye toward - and a need for - automatic 
>>>       
>> conversion"
>>     
>>>  
>>> interestingly enough, if one can generate MAGEv1 to capture the 
>>> details of the microarray experiment, one could also use FuGE to 
>>> export the Material/BioSource individuals as stand alone with the 
>>> ontology annotation and tie them together via the 
>>>       
>> identifier attribute.
>>     
>>>  
>>> "may require a MAGE-ML import into a FuGE DDL database - 
>>>       
>> then export 
>>     
>>> from the database - I'm not clear on this yet"
>>>  
>>> since MAGE-ML has an in-memory model and FuGE does also, then it 
>>> should be just as easy to auto-generate bridging code based on a 
>>> mapping between the two in-memory models as to have to write to a 
>>> database first (which requires the same mapping!).
>>>  
>>> also note that the application/database doesn't have to be 
>>>       
>> based on a 
>>     
>>> FuGE DDL database, it simply needs to be able to import MAGE-ML and 
>>> export FuGE.  i would be out of work if it did.
>>>  
>>> cheers,
>>> michael
>>>  
>>> Michael Miller
>>> Lead Software Developer
>>> Rosetta Biosoftware Business Unit
>>> www.rosettabio.com
>>>
>>>     
>>>       
>> --------------------------------------------------------------
>> ----------
>>     
>>>     *From:* public-semweb-lifesci-request@w3.org
>>>     [mailto:public-semweb-lifesci-request@w3.org] *On Behalf Of
>>>     *William Bug
>>>     *Sent:* Thursday, May 31, 2007 7:36 PM
>>>     *To:* Kei Cheung
>>>     *Cc:* Smith, Barry; Kashyap, Vipul; samwald@gmx.at;
>>>     public-semweb-lifesci@w3.org; 
>>>       
>> obo-relations@lists.sourceforge.net
>>     
>>>     *Subject:* Re: [Obo-relations] Advancing translational research
>>>     with the Semantic Web (Not clear about definition of
>>>     <is_location_of_process>)
>>>
>>>     Hi Kei,
>>>
>>>     Yes - you are right, of course - right now the TGEN 
>>>       
>> infrastructure
>>     
>>>     for the consortium is committed to providing MAGE-ML 
>>>       
>> instances [1].
>>     
>>>     My understanding from speaking with FuGE folks is that the the
>>>     FuGE-stk [2] will provide a means to "convert" MAGE-ML 
>>>       
>> to FuGE-ML
>>     
>>>     (may require a MAGE-ML import into a FuGE DDL database - then
>>>     export from the database - I'm not clear on this yet).  
>>>       
>> Since many
>>     
>>>     of the FuGE model developers were a part of the MGED MAGE model
>>>     development project, there is both an eye toward - and 
>>>       
>> a need for
>>     
>>>     - automatic conversion. As FuGE is intended to cover ALL of
>>>     functional genomics beyond microarray alone, there's a bit more
>>>     abstraction in the data model and some more specific 
>>>       
>> parts of the
>>     
>>>     model will likely not be needed for microarray data.
>>>
>>>     I'm not completely clear on how automatic it will be, but folks
>>>     such as Michael Miller who have contributed to the HCLS IG list
>>>     before would certainly be able to give us the most comprehensive
>>>     answer to that question that is available at this time.
>>>
>>>     One example I found recently is out of the 
>>>       
>> bioinformatics unit at
>>     
>>>     Newcastle U. - the Centre for Integrated System Biology of Aging
>>>     and Nutrition [3] [4].  In addition to being one of the first
>>>     public systems based on the Milestone 3 release of FuGE & the
>>>     FuGE-stk, it has a means of transferring data from the
>>>     ArrayExpress backend - maxdLoad2 [5].  Since the latter 
>>>       
>> system is
>>     
>>>     capable of importing MAGE-ML instances, this provides a 
>>>       
>> route via
>>     
>>>     which one can get from MAGE-ML to FuGE-ML.
>>>
>>>     Of course, we could skip the FuGE step and just look at 
>>>       
>> how to use
>>     
>>>     OBI and other OBO Foundry ontologies to create a SemWebTech
>>>     repository for NIH Neuroscience Microarray Consortium 
>>>       
>> data as is -
>>     
>>>     in MAGE-ML or in the backend model - akin to the ones 
>>>       
>> Alan et al.
>>     
>>>     have set up for the HCLS demo at the NeuroCommons.  We 
>>>       
>> are working
>>     
>>>     with annotating MAGE-based microarray data within MouseBIRN as
>>>     well, so it would be wonderful, if there were some way 
>>>       
>> for this to
>>     
>>>     be included.  One of the goals of what we are doing for 
>>>       
>> microarray
>>     
>>>     data in BIRN is to stay in sync with the consortium in 
>>>       
>> such a way
>>     
>>>     so as to make it possible for us to query consortium data - and
>>>     visa versa.  There are some folks on BIRN whose are also
>>>     associated with the consortium (I believe the Autism groups
>>>     recently added to BIRN are participants of the consortium).
>>>
>>>     Do you know whether others on the consortium - or TGEN itself -
>>>     are working on this task?  We might want to have a call that
>>>     includes some of the core informatics folks in the 
>>>       
>> consortium, in
>>     
>>>     addition to yourself.
>>>
>>>     Cheers,
>>>     Bill
>>>
>>>
>>>     [1] 
>>>       
>> http://arrayconsortium.tgen.org/np2/public/dataAndAnalysisPolicies.jsp
>>     
>>>     [2] http://fuge.sourceforge.net/dev/index.php
>>>     [3] http://www.cisban.ac.uk/cisbanDPI.html
>>>     [4] 
>>>       
>> http://www.cs.ncl.ac.uk/research/pubs/trs/abstract.php?number=1016
>>     
>>>     [5] http://www.cisban.ac.uk/resources.html
>>>
>>>     On May 31, 2007, at 10:05 PM, Kei Cheung wrote:
>>>
>>>       
>>>>     Hi Bill,
>>>>
>>>>     Thanks for describing the evolution of MGED into FuGO. As I
>>>>     understand it, the consortium's microarray data can be exported
>>>>     in MAGE-ML (XML) format. Would it be possible to convert it to
>>>>     the FuGE format?
>>>>
>>>>     Cheers,
>>>>
>>>>     -Kei
>>>>
>>>>     William Bug wrote:
>>>>
>>>>         
>>>>>     Barry beat me to the punch here - 
>>>>>     BUT - 
>>>>>     I would not want to miss out on the specific value of the
>>>>>     proposal Kei has made.
>>>>>
>>>>>     I believe looking closely at how the OBI representation of
>>>>>     microarray-associated instruments, protocols, reagents, data
>>>>>     artifacts, algorithms, etc. - could be put to use in 
>>>>>           
>> describing
>>     
>>>>>     some of the data being produced for the NIH Neuroscience
>>>>>     Microarray Consortium that you are contributing to, 
>>>>>           
>> Kei.  As you
>>     
>>>>>     may already know, many of the experts working on FuGE 
>>>>>           
>> (grown out
>>     
>>>>>     of MAGE which used the MGED Ontology as its shared semantic
>>>>>     framework) are looking for assistance in how to make use of
>>>>>     ontologies when representing microarray data in a FuGE
>>>>>     instance.  As you also probably know, the original
>>>>>     FuGE-associated ontology, FuGO, has expanded its 
>>>>>           
>> domain to cover
>>     
>>>>>     all forms of biomedical investigation (Ontology of Biomedical
>>>>>     Investigation - aka the OBI that Barry cited).  This 
>>>>>           
>> was a part
>>     
>>>>>     of the evolution of FuGO as it began to participate in the OBO
>>>>>     Foundry AND make a commitment to use BOTH the OBO Relations
>>>>>     ontology and BFO.
>>>>>
>>>>>     With that in mind - and considering the NIH Neuroscience
>>>>>     Microarray Consortium is committed to providing array data in
>>>>>     FuGE format - it could be very helpful both to understand how
>>>>>     OBI can be used to provide a formal semantic representation of
>>>>>     important experimental provenance information AND how 
>>>>>           
>> SemWebTech
>>     
>>>>>     in general could be used to provide a more flexible - and
>>>>>     query-able - framework in which to access this 
>>>>>           
>> semantic information.
>>     
>>>>>     Cheers,
>>>>>     Bill
>>>>>
>>>>>     On May 31, 2007, at 9:21 PM, Kei Cheung wrote:
>>>>>
>>>>>           
>>>>>>     Smith, Barry wrote:
>>>>>>
>>>>>>             
>>>>>>>     At 08:52 PM 5/31/2007, Kei Cheung wrote:
>>>>>>>
>>>>>>>               
>>>>>>>>     Hi Barry,
>>>>>>>>
>>>>>>>>     Welcome to the SWHCLS list. Such a discussion reminds me of
>>>>>>>>     the Nature paper: "Are the current ontologies in 
>>>>>>>>                 
>> biology good
>>     
>>>>>>>>     ontologies?"
>>>>>>>>     
>>>>>>>>                 
>> (http://www.nature.com/nbt/journal/v23/n9/full/nbt0905-1095.html).
>>     
>>>>>>>>     The paper uses the MGED (microarray) ontology to illustrate
>>>>>>>>     some of the ontological issues. I'm just curious 
>>>>>>>>                 
>> how the BFO
>>     
>>>>>>>>     principles and practice can help make such a microarray
>>>>>>>>     ontology more ontologically sound and therefore 
>>>>>>>>                 
>> more machine
>>     
>>>>>>>>     readable.
>>>>>>>>                 
>>>>>>>
>>>>>>>
>>>>>>>     We are already working on it:
>>>>>>>
>>>>>>>     http://obi.sf.net
>>>>>>>
>>>>>>>     BS
>>>>>>>
>>>>>>>               
>>>>>>     That's great! I hope we can develop some real use case of it.
>>>>>>
>>>>>>     -Kei
>>>>>>
>>>>>>
>>>>>>             
>>>>>
>>>>>     Bill Bug
>>>>>     Senior Research Analyst/Ontological Engineer
>>>>>
>>>>>     Laboratory for Bioimaging  & Anatomical Informatics
>>>>>     www.neuroterrain.org
>>>>>     Department of Neurobiology & Anatomy
>>>>>     Drexel University College of Medicine
>>>>>     2900 Queen Lane
>>>>>     Philadelphia, PA    19129
>>>>>     215 991 8430 (ph)
>>>>>     610 457 0443 (mobile)
>>>>>     215 843 9367 (fax)
>>>>>
>>>>>
>>>>>     Please Note: I now have a new email - 
>>>>>           
>> William.Bug@DrexelMed.edu
>>     
>>>>>     <mailto:William.Bug@DrexelMed.edu>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>           
>>>>         
>>>     Bill Bug
>>>     Senior Research Analyst/Ontological Engineer
>>>
>>>     Laboratory for Bioimaging  & Anatomical Informatics
>>>     www.neuroterrain.org
>>>     Department of Neurobiology & Anatomy
>>>     Drexel University College of Medicine
>>>     2900 Queen Lane
>>>     Philadelphia, PA    19129
>>>     215 991 8430 (ph)
>>>     610 457 0443 (mobile)
>>>     215 843 9367 (fax)
>>>
>>>
>>>     Please Note: I now have a new email - William.Bug@DrexelMed.edu
>>>     <mailto:William.Bug@DrexelMed.edu>
>>>
>>>
>>>
>>>
>>>       
>>
>>
>>     
>
>
>   

Received on Wednesday, 6 June 2007 20:05:06 UTC