[ome-devel] annotations

[Bernd Jagla]

Hello again,

 

Yet another issue: The INCell analyzer comes with software to analyze the
images it creates. It can analyze the images on the fly and also off-line.
The results are stored (of course) in the same directory structure with
different subfolders for all the analyses. Each analysis file is a text file
with a header section describing the user who performed the analysis, which
module was used and the parameters used. The last part of the file contains
CSV list with a header line and lots of numbers afterwards (some of them
actually mean something).

 

Some of the issues I see are that the only reliable way to link the data is
through the directory structure. The spreadsheet importer needs images IDs
that have to be obtained from OME so a simple script to reformat the results
file wouldn't work.

 

Now I thought of four possible scenarios on how to get this information into
OME:

a)       First run the importer on the "run" file. Within this procedure a
file that knows how to link image ID's with analysis results will be
created. Then execute a script that combines the information. Then use the
spreadsheet importer.

a.       This is too difficult and there are too many steps involved = not a
good solution

b)       Write an importer for the analysis files. This would "somehow" use
the directory structure to identify the screen, plates and wells, i.e.
images to import the results

a.       This has the advantage that the images wouldn't be read in again
(see c)). This "somehow" would mean that the directory would have to be
stored in OME. And I guess it will be a bit unsafe (what happens if the
analysis files resides somewhere where it doesn't belong???

c)       Import the results together with the "run" file.

a.       Probably the best solution, but this would mean that if someone
reanalyzed the images, everything had to be re-imported, images had to be
uncompressed again (this takes a LOT of time).

d)       Lastly after reading about HTML templates I thought one could
create a web-page to link a screen or plate with a specific analysis file

a.       Well here I have absolutely no clue how to do it, but it would fit
my way of thinking in terms of making it easier to link additional
information to already existing data. Maybe one could actually generalize
this to build a structure like for the image importer but this time only for
linked information. Or do you believe that is already done with the
spreadsheet importer???

 

Or, maybe one could combine a) and d)???

 

Please advise!!!

 

Thanks

 

Bernd

 

 

 

Bernd Jagla, PhD
Associate Research Scientist
Columbia University
1150 St. Nicholas Avenue
Russ Berrie Medical Pavilion, Room 520
New York, NY 10032
Tel: 212 - 851 5560 x16
Fax: 212 - 851 5570
e-mail: baj2107 at columbia.edu
  

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.openmicroscopy.org.uk/pipermail/ome-devel/attachments/20051122/f22ec55b/attachment-0001.html