[ome-devel] Questions and proposals

[Ilya Goldberg]

On Nov 9, 2004, at 5:07 PM, Richard Beare wrote:

> Hi Everyone,
>
> I'm still learning my way around OME, so some of my questions are  
> probably going to be silly. I also have proposals that I'd like advice  
> on. I'll start with the stupid questions.
>
> 1) How do you delete things? I can't find any way of deleting images,  
> or datasets. I'm thinking mostly of test scenarios rather than general  
> users.

Try:
ome help admin data delete MEX
It is likely to delete more than you thought it would, so be careful,  
and try the -n flag first.
Only available from CVS so far...

> 2) There seem to be a lot of tables relating to microscope details in  
> the database. I can't find where to enter or query this information.

Right now, the only way to enter it is to read an OME format XML file.   
It hasn't been a priority for a user interface because this kind of  
information need only be entered once, and eventually be entered for  
you by the manufacturer (lets hope!).  Subsequent files (images) will  
simply refer to fields in this structure by reference.

>
> Also, how should I automatically enter information relating to the  
> capture of images. Is the annotation field the only option? Things  
> like xyz stage position would be useful to enter.

CVS has a more generalized mechanism for scanning this type of  
information from filenames.  If this information is in a  
manufacturer-specific file, you will need to modify one of the importer  
scripts to read this information from this type of file.   
Alternatively, this can be done by reading OME XML.  The datatype  
(Semantic Type in OME parlance) to contain this information is defined,  
but probably not in a finalized state.  I believe the DKFZ group is  
working on this.

>
> Ultimately we are going to have a barcode reader identifying slides  
> prior to scanning so this sort of information needs to go somewhere  
> to. How much of this sort of thing is ready to go and how much would  
> require extensions to the semantic types?

Some is ready to go, some will require new STs.  New STs are very easy  
to define, so this is not necessarily a big hurdle.  The big hurdle is  
more sociological because maximum benefit will be achieved by agreeing  
with other people on the structure of this ST so that it can be re-used  
and shared between different applications.

>
> 3) I'm trying to install the cvs version to see whether it corrects  
> problems with tiff importing. I get the error:
>
> ERROR LOADING XML FILE  
> "/home/richardb/OME/src/OME/src/xml/OME/Annotations/ 
> classification.ome" -- OUTPUT: "OME::ImportExport::SemanticTypeImport:  
> While processing Semantic Type CategoryGroup, existing semantic  
> element Name is stored in CATEGORY_GROUPS.NAME instead of  
> CATEGORY_GROUP.NAME.
>
> I haven't chased this yet.

Unfortunately the CVS version is in an interim state right now.  It  
works well off with a 'fresh' database (its tested nightly by an  
automated script), but we are in the middle of a transition in how we  
define categories (from dataset-granularity to global-granularity), so  
your mileage may vary depending on what day you update from CVS.  We  
anticipate this to be resolved in the next week or so.  Please try a  
"latest" version, because a few problems like this were corrected  
recently.  If they weren't we would like to work with you to resolve  
them so that they don't come back to bite us when we finally do our  
next release.

> ----------------------------------------------
>
> The idea I'd like advice on is the inclusion of R in the same way that  
> matlab has been.
>
> From my point of view this is important because R is the engine we use  
> to drive our image analysis. It is also free and could be used for  
> statistical work on the data retrieved from images.
>
> Where should I start looking to think about this work?

Unfortunately, "Analysis Handlers" are fairly complex pieces of code  
because of the generality they entail and the fact that they serve as  
interfaces to external code.  The best way to move forward is to  
examine the existing examples in OME/src/perl2/OME/Analysis/Handlers/

We can certainly help with this kind of integration, but I don't know  
of anyone off-hand in our project that has expertise in R.  The  
MatlabHandler is getting a lot of use and abuse from our group these  
days, so it works pretty well.  I believe that the CLI  
(Command-line-interface) Handler may be broken presently - others  
(Josiah) would know for sure.

>
> Thanks

Glad to have you on-board
-Ilya


>
> _______________________________________________
> ome-devel mailing list
> ome-devel at lists.openmicroscopy.org.uk
> http://lists.openmicroscopy.org.uk/mailman/listinfo/ome-devel
>