This is probably a bit too rigid, but it works for us at the moment. If we plan to scale things up it may be worth properly thinking things through.
-
slide1
-
scene1
- tiles
- round1
- 0.ome.tiff
- ...
- round2
- 0.ome.tiff
- round1
- image.ome.tiff
- image.ome.zarr
- tiles
-
scene2
- ...
-
.brightfield
-
.illumination
- flatfield-{round}.tiff
- darkfield-{round}.tiff
-
.scan-logs
-
.logs
- nextflow
- before-annotate.log
- before-annotate.report.html
- before-annotate.trace.txt
- before-annotate.timeline.html
- after-annotate.log
- after-annotate.report.html
- after-annotate.trace.txt
- after-annotate.timeline.html
- illum-profiles
- make-illum-profiles-{round}.dask-performance.html
- plot-illum-profiles-{round}.svg
- stack-tiles
- stack-tiles-{round}-{scene}.dask-performance.html
- unmixing-mosaics
- make-unmixing-mosaic-{round}.dask-performance.html
- segmentation
- segment-{scene}.dask-performance.html
- nextflow
-
-
calibration-dir
- cube-alignment
- gifs
- images
- plots
- scales-shifts
- .logs
- cube-alignment-{timestamp}.dask-performance.html
- dark
- lamp
- stage-xy
- cube-alignment
- for conda to specify certain functions as entry points, the filenames of the files containing said functions must be importable (e.g. no dashes in filename)