[Data-modeling] Additional properties to Disease/Physician

[Faye Li]

Since it's so specialized, I'm offling this discussion with Dan. If 
anyone is interested in getting involved in the discussion, email me 
directly to get cc'd on.

-- Faye


Dan Ruderman wrote:
> Hi Faye,
>
>   
>> 1) Type: "Disease or medical condition"
>> New property: "Associated Genes", with expected type "Gene"
>> Reverse property on type "Gene" will be "Associated Diseases" (will be 
>> done once I get a Biology domain admin to help)
>>
>> For example, Huntington Disease is associated with the hungtingtin gene:
>> http://sandbox.freebase.com/view/guid/9202a8c04000641f800000000005ead2
>>
>>   
>>     
> Two comments on this.   First, people are more and more
> associating gene groups (say, that belong to a particular
> pathway) with diseases.   So you might want to think about
> an additional property relating diseases to the gene group type.
> Second, though it is a bit more work, it would be more
> scientifically useful to connect the gene and the disease
> though a relationship, like the way a drummer relates to
> a band though band membership.  This way the relationship
> between disease and gene can be annotated (e.g. it was stated
> in these publications, or on this web site).   My concern is
> that what we think is true today may not be true tomorrow,
> so you ideally want to track who is saying what rather than state
> it as fact.  This makes the data model more complicated, but it might
> be worth the hassle.
>
>   
>> Question: Would it be useful to also add a (unique?) property 
>> "Cytogenetic Band" (expected type "Cytogenetic Band") to the Disease 
>> type? And if so, would a reverse property from "Cytogenetic Band" back 
>> to "Disease" make sense? It would be trivial for me to add both. In the 
>> case of "Huntington Disease", the property would point to the "Human 
>> Cytogenetic Band 4p16.3" topic:
>> http://sandbox.freebase.com/view/guid/9202a8c04000641f80000000051dc7bc
>>   
>>     
> For genes implicated in disease, the cytogenetic band can be
> determined from the locus property on the gene itself.  However,
> there are diseases which implicate entire chromosomal regions,
> like cancer (e.g. http://progenetix.net/progenetix/).  But the
> involvement of any given chromosomal region might depend on
> the stage of the cancer, and the particular tumor you are looking
> at.
>
> I'd be happy to discuss any more thoughts on either of these
> issues off-line.
>
> Best,
> Dan
>
> _______________________________________________
> Data-modeling mailing list
> Data-modeling at freebase.com
> http://lists.freebase.com/mailman/listinfo/data-modeling
>
>