Datasetpublic

einarolafsson/spacr-example-import

spaCR — Import test data The same four microscope fields written in every container format and filename convention the Import module of spaCR reads, each with its cell, nucleus and pathogen masks and the measurements of its cells. It is the data behind Load test data… on the Import screen: pick a variant, and spaCR fills the screen with it and previews the import, so you can see every file land on the well, field and channel it came from. About 283 MB in one uncompressed archive… See the full description on the dataset page: https://huggingface.co/datasets/einarolafsson/spacr-example-import.

sourceHugging Facemitupdated 3d agoView on Hugging Face

0likes27downloads

Dataset Card

spaCR — Import test data

The same four microscope fields written in every container format and filename convention the Import module of spaCR reads, each with its cell, nucleus and pathogen masks and the measurements of its cells. It is the data behind Load test data… on the Import screen: pick a variant, and spaCR fills the screen with it and previews the import, so you can see every file land on the well, field and channel it came from.

About 283 MB in one uncompressed archive, spacr-example-import.tar. One download covers every variant.

Where the fields come from

Wells E01 and E02, fields 9 and 10, channels 1–3 of the toxomito example plate `plate1` (`einarolafsson/toxomito`): Toxoplasma-infected host cells, channel 1 nuclei, channel 2 host cells, channel 3 parasites.

—Images are the raw acquisition planes (16-bit, not rescaled), cut to 896 × 896 px. The window is the one that keeps the most whole cells.
—Masks are the cell, nucleus and pathogen label images of the Mask run on the same plate (einarolafsson/spacr-example-measure), cut the same way. Every object the cut went through was removed, so no mask holds half an object.
—Measurements are the Measure rows of exactly the cells left in the masks (einarolafsson/spacr-example-annotate). reference/measurements.db has them as spaCR wrote them (cell, nucleus, pathogen and cytoplasm tables); each variant's measurements.db has the cell table the way a collaborator's table would arrive: spaCR's own key columns removed, and an image column naming that variant's own file.

Channels are named DAPI, FITC and TRITC in the conventions that name channels rather than number them. Import numbers named channels in sorted order, and these three sort in channel order.

The variants

variants/<key>/plate1/ holds the images, variants/<key>/masks/{cell,nucleus,pathogen}/ the masks named the same way, and variants/<key>/measurements.db the table.

Key	Button	Files	Naming (`metadata_type`)	Example
`zeiss_czi`	Test Zeiss CZI import	Zeiss CZI, one plane per file	zeisszensplit_tiles	`E01/plate1_S00001_T00009_C00001.czi`
`cellvoyager`	Test Yokogawa CellVoyager import	TIFF	cellvoyager	`plate1_E01_T0001F009L01A01Z01C01.tif`
`cq1`	Test Yokogawa CQ1 import	TIFF	cq1	`W0097F0009T0001Z000C1.tif`
`opera_phenix`	Test Opera Phenix import	TIFF	opera_phenix	`Images/r05c01f09p01-ch1sk1fk1fl1.tiff`
`imagexpress`	Test ImageXpress import	TIFF	imagexpress	`TimePoint_1/plate1_E01_s9_w1<GUID>.TIF`
`arrayscan`	Test ArrayScan import	TIFF	arrayscan	`plate1_E01f08d0.TIF`
`arrayscan_kinetic`	Test ArrayScan kinetic import	TIFF	arrayscan_kinetic	`plate1i3t001E01f08d0.TIF`
`evos`	Test EVOS import	TIFF	evos	`scan_R_p1_z1_0_E01f09d0.tif`
`incell`	Test IN Cell import	TIFF	incell	`E - 01(fld 9 wv DAPI - DAPI).tif`
`scanr`	Test ScanR import	TIFF	scanr	`data/E1--W00097--P00009--Z00000--T00000--DAPI.tif`
`cytation`	Test Cytation import	TIFF	cytation	`E1_01_1_9_DAPI_001.tif`
`leica_matrix_screener`	Test Leica Matrix Screener import	OME-TIFF, one plane per file	leicamatrixscreener	`slide--S00/chamber--U00--V04/field--X08--Y00/image--L00--S00--U00--V04--J20--E00--O00--X08--Y00--T00--Z00--C00.ome.tif`
`leica_lasx_series`	Test Leica LAS X series import	TIFF	leicalasxseries	`E01/Series009_z00_ch00.tif`
`leica_lasx_series_time`	Test Leica LAS X timelapse import	TIFF	leicalasxseries_time	`E01/Pos008_t000_z00_ch00.tif`
`nikon_nis_xy`	Test Nikon NIS-Elements import	TIFF	nikonnisxy	`E01xy09c1.tif`
`nikon_jobs`	Test Nikon JOBS import	TIFF	nikon_jobs	`WellE01_ChannelDAPI_Seq0009.tif`
`micromanager_mda`	Test Micro-Manager import	TIFF	micromanager_mda	`E01/img_channel000_position008_time000000000_z000.tif`
`zeiss_zen_split_tiles`	Test Zeiss ZEN split tiles import	TIFF	zeisszensplit_tiles	`E01/plate1_S00001_T00009_C00001.tiff`
`custom`	Test custom naming import	TIFF	custom `(?P<plateID>[^_]+)__(?P<wellID>[A-Z]\d{2})__site(?P<fieldID>\d+)__(?P<chanID>[A-Za-z0-9]+)`	`toxo-plate1__E01__site09__DAPI.tif`
`auto`	Test folder-layout import	TIFF	auto	`E01/fov09_ch1.tif`
`ome_tiff`	Test OME-TIFF import	OME-TIFF, three channels per file	auto	`E01/field009.ome.tif`
`tiff_stack`	Test TIFF stack import	ImageJ TIFF stack, three channels per file	auto	`E01/field009.tif`
`nikon_nd2`	Test Nikon ND2 import	Nikon ND2 (public sample)	auto	`A01/WellA01_ChannelBF_Seq0001.nd2`
`leica_lif`	Test Leica LIF import	Leica LIF (public samples)	auto	`A01/FRAP.lif`

What is true of every file: `manifest.csv`

One row per file. path is relative to the unpacked import_example/ folder; well, field, channel, z and t are what the file truly is, and source_field is the field number on the original plate. field_identity says what an import should recover: number means the field number itself (field 9 stays field 9); rank means the order within the well, for the variants whose names carry no field number (a folder layout, a Leica Matrix Screener X/Y position, a file of channels). The manifest is the last member of the archive, so a download that died part-way has none and reads as absent.

Verified

Every variant was imported with spaCR's own planner and each output plane compared with the manifest: well, field, channel, z and t, every mask paired with its field, and every measurement row matched to an object in the masks.

Variant	Result
`zeiss_czi`	passed
`cellvoyager`	passed
`cq1`	passed
`opera_phenix`	passed
`imagexpress`	passed
`arrayscan`	passed
`arrayscan_kinetic`	passed
`evos`	passed
`incell`	passed
`scanr`	passed
`cytation`	passed
`leica_matrix_screener`	passed
`leica_lasx_series`	passed
`leica_lasx_series_time`	passed
`nikon_nis_xy`	passed
`nikon_jobs`	passed
`micromanager_mda`	passed
`zeiss_zen_split_tiles`	passed
`custom`	passed
`auto`	passed
`ome_tiff`	passed
`tiff_stack`	passed
`nikon_nd2`	passed
`leica_lif`	passed

What a pass proves, and what it does not: the files were written to follow each convention as spaCR's convention table documents it (the same table Mask uses, including which numbers count from zero), and read back by the Import module. It shows Import inverts those conventions. It cannot show that the table is right about every instrument; conventions marked provisional in spaCR's Naming list are the ones built from few real examples.

ND2 and LIF

No open library writes Nikon ND2 or Leica LIF, so those two variants are not these fields: they are public sample files from downloads.openmicroscopy.org, redistributed under their licence, CC BY 4.0 (the text is in each folder as LICENSE-CC-BY-4.0.txt). They carry no masks, so their buttons fill the Format Converter rather than the Import screen.

—variants/nikon_nd2/plate1/A01/WellA01_ChannelBF_Seq0001.nd2 — https://downloads.openmicroscopy.org/images/ND2/maxime/BF007.nd2 -- (c) Maxime Woringer, CC BY 4.0
—variants/leica_lif/plate1/A01/FRAP.lif — https://downloads.openmicroscopy.org/images/Leica-LIF/seanwarren/150519FRAPtestROIschromagreen/150519FRAPtestROIschromagreen.lif -- (c) Sean Warren, CC BY 4.0
—variants/leica_lif/plate1/A01/PR2729_frameOrderCombinedScanTypes.lif — https://downloads.openmicroscopy.org/images/Leica-LIF/michael/PR2729_frameOrderCombinedScanTypes.lif -- (c) Michael Goelzer, CC BY 4.0

Using it

In spaCR, open Import and press Load test data…. The first choice downloads the archive into ~/.cache/spacr/example_data/plate1/import_example/; later choices open from there. From a terminal: spacr-download import.

Licence

MIT, like spaCR's other example sets, except the two sample folders variants/nikon_nd2 and variants/leica_lif, which are CC BY 4.0 by the authors named above.