Attending: Petr, Seb, Frances, Mark, Jean-Marie, Simon, Josh, Dominik, June, David, Melissa, Kevin, Jason, Will
Start: 2:00 pm
(2-3 minutes each)
biology studies (idr0070/HDBR atlas, idr0077/flower development via SPIM, idr0079/3D confocal imaging in zebrafish)
Josh: publication as zarr? (Holding off on masks. See below)
might come back to in the future)
ManagedRepository vs client path.
have the discussion as the datasets are published.
Simon: can submitters expect there to be a zarr? No.
Chris: community doesn’t understand that. OME == IDR
JRS: all in good time.
Seb: like the download (aspera) page
Chris: “some datasets are available from Zarr”
JRS: service provision is evolving.
Bug in stack (s3fs/fsspec/dask) when arrays are sparse
Restrictive S3 server. Missing chunk -> access dies.
Probably a classical issue with masks
Three mask options:
Labelled image but doesn’t support overlaps.
6D booleans but 8x storage (Better as probability mask)
One array per mask but 8x storage and lots of groups
Simon: first idr study with masks was overlapping
Emil: case of multiple segmentation with varying parameters
Storing each “segmentation” as a separate group?
Simon: need to deal with multiple groups in all cases?
groups. Avoid 150K groups
Summary: smallish number of non-overlapping labelled images
incoming OME mail.
Report any weirdnesses asap.
Can then start heavily on the thumbnail front.
in the wild.
JRS: people asking about how OME2020 was done.
How were videos done.
Breakouts for fun.
(5 min. max; tech. Discussion should be highlighted to relevant people and rescheduled)
(20-25 minutes plus 15 minutes questions max)