Attending: Will, Petr, Jason, Jean-Marie, Mark, June, Wilma, Simon, Andreas, Kevin, Melissa, Josh, David, Erin, Emil, Seb (14:17)
Start: 2:00 pm
(2-3 minutes each)
New data upload mechanism:
IDR 0.8.5 released: two new 3D mouse studies
IDR 0.8.6: WSI data: a new histopathology study with lung cancer
images + 1000 new HPA antibodies
S3 fun (CORS, public, private, clients, etc.)
Simon: using Amazon JS viewer to browse idr-upload
Need a web upload client via minio
Pushing on the final breaking changes (“labels/”)
Next steps: handle multi-resolution zarr.
Use cases, drivers, etc.
JM: Cellprofiler (python2) issues with zarr - need to test again
Jason: will provide datasets to test, with masks.
deep copy (Mark)
a situation where hierarchies belong to different users (at least in R-O and R-A). The webclient, however, allows you to create such hierarchies by dragging things from one container to another. The discrepancy leads to links being cut.
understood it. Instead:
people are shutting off. They don’t want to hear “link is cut”. → “Why?!” (They don’t care about R-O or R-A) The more complex the rules the deeper the hole we are digging for ourselves. Intuitive expectation: “everything will stay the same except what I want to move.“
Jason: links shouldn’t be cut?
Petr: exactly! NO links should ever be cut.
Mark: received wisdom was P/D/I groups were ‘same owner’
fix it: unlink, deep-copy
change ownership, before running the chown command
anything else? (component releases?)
SA: Josh: pgslave is growing
n5/zarr looking at open-cv for downsampling
Tile-db instead of pytables?
Omero microservices for 3D analysis, imput from s3 buckets
Josh: what do you want to try
Emil: labelled masks, WSI. Can convert slides and create masks
CLA and Code of Conduct in Draft
Sanger proposal funded - single cell imaging/analysis
(5 min. max; tech. Discussion should be highlighted to relevant people and rescheduled)
(20-25 minutes plus 15 minutes questions max)