[BioC] GEOmetadb and gse organism type

[Wacek Kusnierczyk]

Jack Zhu wrote:
> Hi Wacek,
>
> Please see my answer to the web interface issue below.  Please let me know
> if you have any questions.
>
>
>   
>> another issue:  as far as i can see, the web interface to geometadb
>> allows one to filter data series by organism, though it's not clear to
>> me whether this means platform organism, sample organism, both, or any.
>> it might be good to clarify this.
>>     
>
> As you can see, on the web interface of the GEOmetadb application (
> http://gbnci.abcc.ncifcrf.gov/geo/geo_search.php ) you can search for anyone
> of the three GEO entities (GPL, GSE and GSM) with any combination of terms
> from any of the three.  The design provides an effective way, in my opinion,
> to do crossing GEO type search, which usually needs a complex SQL join
> command to achieve from command line. It allows you search GSE with an
> organism name.  

thanks for the explanation. 

yes, i know this search page.  my point was, when i choose, say, 'homo
sapiens' in the 'organism' field while searching for gse entries, what
does it mean, exactly?  which organism fields are examined, and what's
the boolean combination?  maybe it's explained there on the page, i just
haven't found it?


> It also includes MySQL feature, Fulltext serach, in it which
> allows Boolean mode.   

hmm, is this working properly?  the following query:

    drosha

is assumed to return all entries with 'drosha' -- there is just one
(with organism = homo sapiens).  that's fine;  the following query:

    mirna

is supposed to return all entries with 'mirna' -- there are 17.  the
following query:

    mirna drosha

is supposed to return all entries with 'drosha', 'mirna', or both --
there are 18 such entries (no overlap).  the following query:

    mirna +drosha

is supposed to return all entries with or without 'mirna', but
necessarily with 'drosha' (the mysql doc page you're referring to says
"A leading plus sign indicates that this word /must/ be present in each
row that is returned" and "By default (when neither |+| nor |-| is
specified) the word is optional, but the rows that contain it are rated
higher").  but this query returns 18 entries, while only one of them
contains 'drosha'!

as far as i can see, the same happens with any selection of keywords
prefixed with '+', they work in the optional mode irrespectively of the
prefix.  (but the exclusion prefix '-' seems to work.)


> Behind the scene is a complex SQL join expression,
> the same as Wacek wrote.
>   

you mean i reproduced the very same query you're using?  interesting.

vQ