[Orechem] Modeling Attributes

[Prasenjit Mitra]

Hi all,

I may not be there at today's phone meeting, but, I wanted to discuss the
following.  I think Karl Mueller will be there from Penn State and I will
talk to him afterwards to find out what was discussed.  I am giving some
of my naïve/preliminary thoughts below.

In general, it does not make sense to talk about classes and properties
separately.  We should specify in the model which classes have which 
properties. 
 
By looking at the queries and the data model, I realize that we cannot 
answer most of the queries if with the preliminary data model that we have.

Query 1:  [Show me all compounds ... bind ... sub-structure]
Are we entering the sub-structure via drawing or via SMILES, etc.?
Is there a program that can be called to find out what substructures bind to
what compounds?  Or do we need to implement it?

Query 2: [ ... all papers ... compounds ... bind ... sub-structure]
We run 1, then, find the answers and try to find those compounds in the 
papers.  Can be done more optimally than that, perhaps.  If Query 1 can be
answered, this should be easy.

Query 3: [ ... spectral data ... acetic acid][NMR spectral data ... acetic acid ... Brucker spectrometer]
We need each dataset to have the type of data they are associated with.
** Add attribute to dataset class **.
To answer the last part, we need to ** add attribute "equipment-type" to 
the dataset class. And attribute "methods" or "experiment type" to dataset class.** 

I did not get exactly what the query was and what the expected answers were
in this part.  Can someone please elaborate it a bit more on what exactly
the queries are and what should be the expected answers?

" Entities required in the model (most are covered by Carl's previous examples):

chemical reactivity maps or tables (reactome ?)
the reactome to me means a table of compounds (probably with thermodynamic and - more difficult - kinetic rate data) that have known or predicted chemical reactivity with a compound (or sub structure)

tables of spectral data (peak lists with intensities) - specific datasets
correlations in multidimensional spectra - more specific datasets
experimental chemical methods and instrumentation
computational chemical methods, programs/environments, and instrumentation"

We should start drawing class-attribute graphs/hierarchies and editing them
so that we have some starting point for the face to face meetings at the
end of this month.

Best Regards,
Prasenjit