Subcellular

Nuclear & RNA Foci

Count the RNA foci or DNA-damage foci in every nucleus of your fluorescence images, and compare them across samples and conditions.

ConductVision segments nuclei, finds the foci inside each one, assigns every focus to its nucleus and exports counts, intensity and area per nucleus.

Request a Demo Back to Applications

Immunofluorescence micrograph: red DNA-damage (53BP1) foci as discrete puncta inside blue-stained cell nuclei, with the green tubulin cytoskeleton around them. — Adapted from Velichko Artem, Wikimedia Commons · CC BY 4.0

Per nucleus

Foci readout

Count + intensity

& area

Per condition

Batch export

Outputs

What gets measured

Every metric is reported per nucleus and for the whole population.

Per nucleus

Foci count
Mean foci intensity
Total foci area

Population

Foci-positive fraction
Foci-per-nucleus distribution

Signal

Integrated intensity
Background-corrected

Screening

Per-well aggregate
Dose-response ready

Workflow

How It Works

Load

Load a fluorescence image or a folder of dose points.

Segment nuclei

AI separates touching nuclei to define each per-cell region.

Detect foci

Intranuclear puncta are detected and assigned to their parent nucleus.

Export

Download per-nucleus foci counts, intensities, and per-well aggregates.

Accuracy

Spot detection scored against expert annotations

We scored a Laplacian-of-Gaussian spot detector on a public set of 446 γ-H2AX images of irradiated human lymphocyte nuclei, where γ-H2AX marks DNA double-strand breaks.¹ Each score comes from 90 held-out images that played no part in setting the detector.

F1, detections against expert annotations: 0.79
Precision, the share of detections that were annotated foci: 0.84
Recall, the share of annotated foci found: 0.75
Agreement of counts per image (ICC): 0.78

Limits of the scores

The dataset's annotators marked only large, bright foci and left out small or dim ones.¹
The scored foci are DNA-damage foci in irradiated lymphocytes.
The detector has no score on RNA foci.

DM1 RNA foci

Foci per nucleus in published DM1 images

In myotonic dystrophy type 1 (DM1), RNA from an expanded CTG repeat in the DMPK gene stays in the nucleus as foci that sequester MBNL splicing proteins.² In neurons converted directly from DM1 skin fibroblasts, most nuclei carried CUG repeat RNA foci.² Counting by hand across 324 nuclei per sample, Eltahir and colleagues found 3.82 foci per DM1 nucleus.² In the same study, actinomycin D, a potent transcription inhibitor, cut foci per nucleus by 56% at 100 nM and 66% at 200 nM.^2,3

In a 2D cell-image check, we ran the spot detector without tuning on one published image per condition from that study.² The settings used here differ from the scored ones and have no score of their own.

Foci per nucleus found by the spot detector in one published field per condition
Neurons	Nuclei	Foci per nucleus	95% interval
Control	9	0.00	0.00 to 0.00
DM1	16	1.94	0.81 to 3.81
DM1, 100 nM actinomycin D	7	1.00	0.43 to 1.57
DM1, 200 nM actinomycin D	7	0.57	0.14 to 0.86

What the check shows, and its limits

The detector undercounts, finding 1.94 foci per DM1 nucleus where the authors counted 3.82 by hand.²
Each row is one published field of 7 to 16 nuclei, so its interval describes that field and leaves out variation between experiments.
On these single fields, the detector found 48% fewer foci per nucleus at 100 nM and 71% fewer at 200 nM, against the authors' 56% and 66%.²
The 95% interval of the 100 nM change runs from 45% more foci to 81% fewer, so it crosses zero; at 200 nM it runs from 17% to 91% fewer.
Eltahir and colleagues saw some cell toxicity at 200 nM.²

Four fluorescence fields of neurons made from skin fibroblasts, nuclei blue (DAPI) and CUG repeat RNA foci red (Texas red CAG probe). Control neurons show no foci, DM1 neurons many, and DM1 neurons after 100 nM or 200 nM actinomycin D fewer. — Adapted from Eltahir et al. 2022 · CC BY 4.0

Four fluorescence fields of DM1 fibroblasts, nuclei blue (Hoechst) and repeat RNA foci green (Cy3-(CAG)10 probe). Foci show in many nuclei after DMSO, and few or none after Ro 31-8220, gemcitabine or chromomycin A3. — Adapted from Ketley et al. 2014 · CC BY 3.0

A high-content screen of DM1 and DM2 fibroblast lines found compounds that cleared nuclear foci.⁴ On the screen's published DM1 images, the spot detector counted 21 foci after DMSO, 1 after Ro 31-8220 and none after gemcitabine or chromomycin A3.⁴ Chromomycin A3 binds G-C base pairs and inhibits RNA synthesis.⁴ For gemcitabine and hypericin, toxicity tracked the loss of foci.⁴ Even where foci disappeared, the repeat RNA stayed in the nucleus.⁴

So a lower foci count, on its own, cannot separate a rescue from toxicity or blocked transcription. In a pilot, ConductVision reports foci per nucleus beside the share of MBNL1 inside foci and the nuclei left in each field. Splicing comes from the lab's own RT-PCR. The pilot reports a rescue when foci and MBNL1 sequestration fall, splicing improves and the count of nuclei holds.

Repeat expansion disorders in cells and organoids

Finding foci in 3D

To test spot detection in 3D, we simulated image volumes, each holding 6 nuclei with 2 to 6 foci at known positions. On 8 volumes held out from setting the detector, it found 195 of the 198 foci with no false detections (F1 0.992). In further simulated volumes at the same setting, F1 stayed at 0.943 or above for foci whose peak brightness was at least 12 times the noise, and fell to 0.40 at 4 times. All of these volumes are synthetic, so they test the 3D detector and leave staining, optics and real nuclei to a pilot.

Corneal disease

RNA foci in Fuchs endothelial corneal dystrophy

Fuchs endothelial corneal dystrophy affects about 5% of middle-aged Caucasians in the United States.⁵ Its strongest genetic link is an expanded CTG repeat in an intron of the TCF4 gene, found in most people with the disease.⁵ In corneal endothelial cells the repeat RNA forms nuclear foci that hold the splicing factor MBNL1, as in DM1.⁵

In endothelium from 8 people with the expansion, 33% to 88% of cells carried foci, and the 5 samples without it had none.⁶ Hu and colleagues found about 2 sense-strand foci per endothelial cell.⁷ Each focus was a single RNA molecule.⁷ In corneal endothelial cells grown from affected donors, an antisense oligonucleotide against the repeat cut foci, MBNL1 recruitment to foci and missplicing.⁸

Single-molecule foci can be dimmer than the large, bright foci in the benchmark, so a pilot begins by checking ConductVision's counts against the lab's own counts by hand. The pilot then reports foci per nucleus and the share of cells with foci in each sample.