2014-03-04 Tuesday Team Meeting
Dundee: Simon, Chris, Kenny, Petr, Emil, Mark, Jean-Marie, Blazej, Gus, Colin, Balaji, Andrew, Dominik, June, Will Remote: Douglas, Kelli, Mark H, Sebastien S, Sebastien B, Andreas, Ian, Roger, Curtis, Melissa, Josh B, Carlos
Agenda - 2:30pm Start
Accepting minutes from last meeting
- accepted
Project Status/Testing Update - 14:31 UK
-
Petr: working on permissions testing, concentrating on move
- creating tables of permissions focusing on specific combinations matching stories
- not sure what should happen, Balaji helping to determine what does happen
- checking that client and server behavior match
- move stories include various objects annotated
- Jean-Marie: considering further format testing
- Mark: will try to match permissions tables against ongoing graph traversal work
- Mark: will try to understand BasicACLVoter and explain to Petr
- Gus: report of ROI intensity analysis slower if drawn on 5.0
-
Jean-Marie: work on evaluating client on iPad and Android
- Gus has started evaluating iPad
-
Petr: deployment of Squash is surprisingly easy
- Kenny: will work on it
- Jean-Marie: intent to release consortium tools two weeks after 5.0.0
-
Petr: working on permissions testing, concentrating on move
Glencoe Update - 14:38 UK
- Josh B: search stuff being rebased from dev_5_0 to dev_5_1
- Josh B: looking at rebasing firehose
-
Seb B: Carlos opened PR against dev_4_4: intentional?
- Carlos: yes, should also go against 5, easy rebase
- Seb B: will talk to Kenny, may reactivate some dev_4_4 CI jobs
- Jean-Marie: will then be in if we have to do another 4.4 release
- Will/Simon: worth the effort to test at this time?
- Jean-Marie: could leave dev_4_4 PR open for now
- Carlos: could open dev_5_0 PR for testing
Consortium Update (every month) - 14:43 UK
-
Seb B: meeting in Edinburgh
-
progress with integrating clients with OMERO.insight
- FLIMfit, Michael’s tools
-
FLIMfit, U-track now available from downloads site
- OMERO.figure should also live there
- also could do OMERO.searcher, webtagging, Michael’s tools
-
many ready to go, or nearly so, main pending question is Michael’s tools, including its name
- Jean-Marie: can talk to Michael
- Jean-Marie: Applications trello board
-
progress with integrating clients with OMERO.insight
- Jean-Marie: using breaking tag from Seb B, now a couple of CI jobs
-
Ian: in this release do we plan to harmonize Matlab versions?
- Chris: no
- Jean-Marie: indicate how to switch between MCR, as it is tricky
- Jean-Marie: in longer term, version in Dundee will be used for testing
- Jean-Marie: must decide on parameters
-
Seb B: meeting in Edinburgh
Upcoming meetings/conferences/outreach visits - 14:52 UK
- Will: Alex in Sussex requests postponing visit, now discussing
- Jean-Marie: Helen submitted abstract for Manchester meeting following workshop
-
Gus: submitting poster for Crieff meeting
- Jean-Marie: can check poster over and print
- Chris: note that printing service will be busy just beforehand
Simon Li: OMERO.searcher update - 14:55 UK
-
image-based search
-
currently optimized for sub-cellular localization, from CMU’s work (Murphy Lab)
- similarity based on distribution within cell
- features calculated with script and stored outside OMERO DB as OMERO.tables is too slow
- images compared pairwise with distance metric
-
search results ordered by degree of similarity
- scores are useful for relative ranking, not significant as absolute values or ratios
- spreadsheet export includes rank column
-
currently optimized for sub-cellular localization, from CMU’s work (Murphy Lab)
- proof of concept, intended to be easy to use, not aimed at machine learning experts
- user interface is prototype, plenty of review yet to be done
- Jean-Marie: CMU develops analysis algorithms, Simon integrates into OMERO
-
Balaji: what confidence have we that the search has useful accuracy?
- Simon: works well on CMU’s dataset, probably will on ones that look similar
- not scale-invariant, not obvious how to approach differently sized images
-
Jean-Marie: OMERO.tables issues?
- Simon: partly speed in transferring much data over Ice, with data conversions, etc.
-
ongoing discussions with Lee K, et al.
- API and generality over analysis algorithms / feature representation and storage, e.g. hstore columns
- Balaji: some existing relevant work from EMBL on PCA to determine relevant features of microscopy images
-
image-based search
Any other business (<5 mins) - 15:29 UK
- Will: may show public-facing gallery web app
Done 15:30 UK