iFISH chromosome spotting (IDR idr0123)

Seven-channel DNA FISH microscopy of human cell nuclei, from the Image Data Resource study idr0123 (Mota et al., Scientific Data 2022). This is the complete Dataset 5 of that study ("iFISH chr1 spotting, replicate 1"): 50 fields of hTERT RPE-1 nuclei in which 23 loci spaced 10 Mb apart along chromosome 1 are labelled with six spectrally distinct fluorophores, plus a DAPI counterstain. The authors' sub-pixel 3D spot coordinates are included as bounding-box annotations, one class per fluorophore.

This is the kind of data general-purpose annotation tools cannot show usefully: every channel needs its own window and level, and the signal you care about is a handful of sub-micron spots per fluorophore. Open any field to see all seven channels as independent layers: toggle fluorophores on and off, drag each channel's window and level, and check the detected spots against the raw signal.

Contents

Fields	50 (fields 001 to 050, all of Dataset 5)
Channels per field	7: DAPI, AF488, AT542, AF594, AT647N, AF700, AF790
Image size	1024 x 1024 px, 16-bit, 0.13 um per pixel (maximum-intensity projection of 61 Z slices, 0.2 um apart)
Format	One classic TIFF per field, one page per channel, OME-XML channel names in the header
Annotations	18,754 FISH spot boxes (9 x 9 px, centred on the published dot coordinates) on 49 fields, plus 39 nucleus polygons on fields 001 to 003
Classes	6 spot classes (AF488, AT542, AF594, AT647N, AF700, AF790) and nucleus
Dataset	fields
Split	None. A single biological replicate; field 046 has no detected spots in the source table
Imaging	Widefield epifluorescence, hTERT RPE-1 cells, combinatorial multi-colour iFISH targeting 23 chromosome 1 loci

Each spot class corresponds to one fluorophore and, in this experiment, to one genomic span on chromosome 1 (from the study's BED file): AF594 chr1:18.2-46.9 Mb, AT647N chr1:56.3-84.9 Mb, AT542 chr1:94.8-120.5 Mb, AF488 chr1:142.6-161.9 Mb, AF790 chr1:171.7-200.3 Mb, AF700 chr1:209.7-238.9 Mb. Each span holds 3 or 4 of the 23 loci.

Using it

Browse a field to explore the channels and the spot boxes. To download everything (free account required):

• ZIP: use the Download button on this project.
• Python SDK: pip install datatorch, then:

datatorch login
datatorch pull datatorch/ifish-chromosome-spotting-idr0123

Annotations export in COCO format through the project's export schema.

Conversion notes

Images. IDR ships each field as seven single-channel Z-stacks of 61 slices (0.2 um steps), 122 MB each. The files here are maximum-intensity projections of each stack, one projection per channel, written as uncompressed 16-bit pages so the full dynamic range survives. Channel names come from the study's OME companion metadata (a488 = AF488, tmr = AT542, a594 = AF594, Cy5 = AT647N, a700 = AF700, ir800 = AF790). Fields 004 to 050 were built with the same code path as the first three; rebuilding 001 to 003 from source reproduced the existing files pixel for pixel.

Spots. Boxes come from the study's Dataset5_Dots_3Dcoordinates_iFISH_chr1.csv (DOTTER output, 18,754 rows), published alongside the images in the IDR metadata repository. Two things were established by inspection rather than taken from the column names. First, the column named x is the image row (vertical) index and y is the column index: sampling the projections at the published positions gives spot-level intensities only with the axes swapped, and background-level intensities as written. Second, the coordinates are 1-based: shifting by -1 px gives the brightest sampled intensity. Each dot is therefore drawn as a 9 x 9 px box centred on (y - 1, x - 1) on the field's projection; the Z coordinate collapses onto the projection. The table's per-dot fields (nucleus ID, Z position, allele label 1/2 or 0 for unassigned, intensity, FWHM, SNR) are not carried into the boxes; take them from the source CSV. Field 046 has no rows in the table. The study's BED file assigns dots to genomic spans; 30 of its 16,070 rows are exact duplicates, and the spans resolve to one per fluorophore, so they are recorded per class above rather than per box.

Nuclei. The 39 nucleus polygons on fields 001 to 003 were drawn in DataTorch when the project was first seeded; they are not part of the IDR release.

Display. The IDR rendering settings for this dataset use red (AF488), green (AT542), yellow (AF594), cyan (AT647N), magenta (AF700), blue (AF790) and white (DAPI); the spot classes use the same colours.

License and citation

CC BY 4.0. If you use this data, please cite:

Mota A, Schweitzer M, Wernersson E, Crosetto N, Bienko M. Simultaneous visualization of DNA loci in single cells by combinatorial multi-color iFISH. Scientific Data 9, 47 (2022). https://doi.org/10.1038/s41597-022-01139-2

Data: Image Data Resource, https://doi.org/10.17867/10000175. Original Z-stacks: idr0123 on the IDR FTP mirror.

Related

• More microscopy datasets on DataTorch
• All public datasets