datatorch

/

MCF7 and A549 NF-kB translocation (BBBC014)

Public

Updated 8 hours ago

microscopyhigh-content-screening

Files

About

MCF7 and A549 NF-kB translocation (BBBC014)

A cytoplasm-to-nucleus translocation assay from the Broad Bioimage Benchmark Collection (BBBC014v1), contributed by Ilya Ravkin. One 96-well plate of MCF7 (human breast adenocarcinoma) and A549 (human alveolar basal epithelial) cells: the transcription factor NF-kB sits in the cytoplasm of resting cells and moves into the nucleus in response to TNF-alpha, so the plate carries a TNF-alpha dose series for each cell line. Nuclei are counterstained with DAPI and NF-kB is stained with FITC. Images were acquired at Vitra Bioscience on the CellCard reader at 10x, one field per well, one image per channel. The source ships the images as a flat folder plus platemap files; here the platemap is attached to every image as file metadata and as an image-level label, so wells can be filtered by cell line, dose and control role in the browser. It is the companion of BBBC013, the other translocation plate from the same contributor, and is laid out the same way.

Contents

Images192 (96 wells x 2 channels, 1360 x 1024, 8-bit grayscale PNG, converted losslessly from the BMPs in the source BBBC014_v1_images.zip)
Annotations192 image-level labels (one per file, the well's control role); no region annotations, the source has no outlines
Classes3: negative-control (16 files, 8 wells), positive-control (16 files, 8 wells), dose (160 files, 80 wells)
Datasettranslocation-bbbc014-v1
SplitNone. One plate: rows A-D are four replicate 12-point TNF-alpha series on MCF7, rows E-H four replicate 12-point series on A549
ImagingFluorescence, 10x; channel 1 NF-kB (FITC), channel 2 DNA (DAPI)

Plate layout

ColumnTNF-alpha (platemap value, g/mL)TNF-alpha (ng/mL)Label
110^-7100positive-control (top dose)
210^-7.52330dose
310^-810dose
410^-8.5233dose
510^-91dose
610^-9.5230.3dose
710^-100.1dose
810^-10.5230.03dose
910^-110.01dose
1010^-11.5230.003dose
1110^-120.001dose
1210^-13 (stands for zero)0negative-control (no TNF-alpha)

The layout is the same in all eight rows; rows A-D are MCF7 and rows E-H are A549.

Files, metadata and channel pairing

Files are named Channel<c>-<well index 01-96>-<row>-<column>-<field>.png under images/<control role>/. The two channels of a well are two separate files (no merged image was made) with the same name apart from the Channel1/Channel2 prefix, in the same folder; each file's pair_file and pair_path metadata point at its partner.

Metadata keys on every file: well, row, column, well_index, field, channel, stain, cell_line (MCF7 or A549, by row), control_role, treatment (TNF-alpha or none), dose, dose_unit (ng/mL), platemap_dose (the raw value from BBBC014_v1_platemap_all.txt), source_file (the original BMP name), pair_file, pair_path. Wells that received TNF-alpha also carry dose_log10_g_per_ml (the value from the platemap spreadsheet); it is left out on the zero-dose wells.

Using it

Browse any image to see its label and metadata. 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/translocation-bbbc014

Annotations export in COCO format through the project's export schema; the labels are per image, not per region, so exports carry categories and image metadata but no boxes or segmentations.

Conversion notes

  • Images come from BBBC014_v1_images.zip (126,643,647 bytes, 192 files). Every source file was opened and checked: all 192 are 1360 x 1024, 8-bit grayscale, uncompressed BMP. Each was re-encoded as an 8-bit grayscale PNG and decoded again to confirm the pixels are identical to the BMP. No resizing, normalisation or channel merging was applied.
  • Files were renamed. The source names look like Channel 1-29-C-05-00.Bmp, with a space and a mixed-case extension; here that file is Channel1-29-C-05-00.png. The original name is kept in source_file.
  • Checks: no two images have identical pixels (SHA-256); every well has exactly one file per channel; the well index in each filename equals the row-major well number; every file sits in exactly one class folder.
  • Channel identity was decided by inspection. The source says each well has a DAPI image and a FITC image but does not say which is Channel 1. Channel 2 shows compact nuclei in every well. Channel 1 shows whole cells with dark nuclei in column 12 and bright nuclei in column 1, which is the NF-kB signal. As a plate-wide check, the pixel correlation between the two channels of a well falls steadily from about 0.86 in column 1 to about 0.53 in column 10, in every row, which also confirms that the dose runs high to low from column 1 to column 12.
  • The platemap text files contain no well identifiers, only dose values in row-major well order (A01 ... H12), separated by bare carriage returns. They were decoded by position and agree with the DoseKey sheet of BBBC014_v1_platemap.xls, which lists row, column, dose and cell type explicitly for all 96 wells.
  • Doses in the source are log10 of the TNF-alpha concentration in g/mL. dose is the same value converted to ng/mL and rounded to three significant figures. The spreadsheet notes that -13 was entered instead of zero concentration, so column 12 has dose = 0 and treatment = none.
  • Control roles are not declared on the source page or in the platemap files; the plate has no separate control wells. The labels here follow the usual convention for a dose series: the zero-dose column 12 is negative-control and the top-dose column 1 is positive-control. Use dose and cell_line if you want a different grouping.
  • Discrepancy: the PlateLayout sheet of the spreadsheet shows 9.523 (no minus sign) in column 6, while its DoseKey sheet and both text files give -9.523. This dataset uses -9.523.
  • Discrepancy: on the source page the two example thumbnails appear to be swapped (the one captioned Positive is the file CNT_Vitra_neg.jpg and the one captioned Negative is CNT_Vitra_pos.jpg). Nothing here depends on those thumbnails.
  • The label import left two small COCO JSON files under imports/ in the project storage. They are not part of the dataset translocation-bbbc014-v1.

License and citation

CC BY 3.0 Unported. The source page states that the SBS Bioimage CNT images are licensed under a Creative Commons Attribution 3.0 Unported License by Ilya Ravkin (license badge and statement on the page). If you use this data, please cite:

We used image set BBBC014v1 provided by Ilya Ravkin, available from the Broad Bioimage Benchmark Collection [Ljosa et al., Nature Methods, 2012].

Source: BBBC014v1, Broad Bioimage Benchmark Collection. Original context: Ilya Ravkin's Invitation to Participate in the Comparison of Image Analysis Algorithms for Intracellular Screening. The images here are format-converted copies of the originals, as described above.

Related

More public datasets

All in microscopy
  • Nuclei Segmentation (DSB2018)datatorch/nuclei-segmentation-dsb2018
  • iFISH chromosome spotting (IDR idr0123)datatorch/ifish-chromosome-spotting-idr0123
  • C. elegans Worm Segmentation (BBBC010)datatorch/worm-segmentation-bbbc010
  • Deep-Fish-Bone zebrafish larval skeleton segmentationdatatorch/deep-fish-bone-zebrafish
  • AxonPath Myelinated Nerve Fibre Segmentationdatatorch/axonpath-myelinated-fibres

Description

Cytoplasm-to-nucleus translocation assay from the Broad Bioimage Benchmark Collection (BBBC014v1). 192 fluorescence microscopy images (1360 x 1024, 8-bit) from one 96-well plate imaged at 10x: 96 wells x 2 channels (channel 1 NF-kB stained with FITC, channel 2 DNA stained with DAPI), one field per well. Rows A-D hold MCF7 cells and rows E-H hold A549 cells, each with four replicate 12-point TNF-alpha dose series (100 ng/mL down to 0.001 ng/mL, plus zero). Ground truth is the platemap: well, cell line and TNF-alpha dose are stored as file metadata on every image, and the control role is also an image-level label (negative-control, positive-control, dose). The two channels of each well are separate files paired through metadata and folder path; there are no cell outlines. License: CC BY 3.0 (Ilya Ravkin). Source: https://bbbc.broadinstitute.org/BBBC014. Please cite: We used image set BBBC014v1 provided by Ilya Ravkin, available from the Broad Bioimage Benchmark Collection [Ljosa et al., Nature Methods, 2012].

Members

0
View all members →

Progress

99% labeled

192 of 194 files

Size

95.6 MB

Datasets

1

Labels

3

Annotations

192