End-to-end pipeline for segmentation and quantitative analysis of fluorescence microscopy data to study phase separation.

---

Quickstart (TL;DR)

For a minimal, end-to-end example of how to run the pipeline, use the commands below:

# 1. Convert ND2 → OME-TIFF (z = 8)
cd nucleus/preprocessing/
python nd2_to_ome_tif.py

# 2. Generate masks
cd ../mask_creation/
python evaluate_nucleus.py --input /path/to/ome/tifs --outdir /path/to/nucleus_masks --gpu --diameter 200
python evaluate_puncta.py  --input /path/to/ome/tifs --outdir /path/to/puncta_masks  --gpu --diameter 20

# 3. Detect puncta per nucleus + aggregate data
cd ../puncta_detection/
python mean_intensity_and_puncta.py \
  --nuc-dir /path/to/nucleus_masks \
  --puncta-dir /path/to/puncta_masks \
  --intensity-dir /path/to/ome/tifs \
  --out-csv /path/to/output.csv \
  --make-triptychs \
  --triptych-out-dir /path/to/triptychs \
  --intensity-channel 2 \
  --puncta-channel 1

Then open csat_estimation.ipynb, set CSV_PATH to the generated CSV, and run the notebook to estimate C_sat.

---

Repository structure

This repository is organized into two main folders:

• nucleus/ – primary focus of the current workflow
  Contains preprocessing, segmentation, puncta detection, and C_sat estimation scripts for nuclear analysis.

• cytoplasm/ – work in progress / older scripts
  This folder is currently not actively used. The latest and maintained pipeline lives under nucleus/.

---

Overall workflow

This pipeline processes raw microscopy data to detect puncta formation and estimate phase separation behavior at differing protein concentrations.

High-level steps:

1. Convert ND2 files to OME-TIFFs (one image per position, fixed z-slice).
2. Segment nuclei and puncta using Cellpose.
3. Determine which nuclei contain puncta.
4. Fit a logistic curve relating nuclear intensity to puncta presence to estimate C_sat.

---

Step-by-step instructions

1. Preprocessing: ND2 → OME-TIFF

If you are starting from an ND2 file:

Navigate to:

cd nucleus/preprocessing/

Run:

python nd2_to_ome_tif.py

Inside nd2_to_ome_tif.py, specify:

• Path to the input ND2 file
• Desired output directory

This script generates OME-TIFF files per XY position, extracting a fixed z-slice (z = 8).

---

2. Mask generation (nucleus and puncta)

Navigate to:

cd nucleus/mask_creation/

Puncta segmentation

python evaluate_puncta.py \
  --input /path/to/ome/tifs/folder \
  --outdir /path/to/desired/output/folder \
  --gpu \
  --diameter 20

Nucleus segmentation

python evaluate_nucleus.py \
  --input /path/to/ome/tifs/folder \
  --outdir /path/to/desired/output/folder \
  --gpu \
  --diameter 200

These scripts use Cellpose 3 to generate segmentation masks for puncta and nuclei respectively.

Channel assumptions:
By default, puncta segmentation uses channel index 1 (second channel) and nucleus segmentation uses channel index 2 (third channel).

For datasets with a different channel ordering, you can explicitly set the channel using the --channel-index argument when running the segmentation scripts.

---

3. Puncta detection per nucleus and data aggregation

Navigate to:

cd nucleus/puncta_detection/

Run:

python mean_intensity_and_puncta.py \
  --nuc-dir /path/to/nucleus/masks/folder \
  --puncta-dir /path/to/puncta/masks/folder \
  --intensity-dir /path/to/ome/tifs/folder \
  --out-csv /path/to/desired/output.csv \
  --make-triptychs \
  --triptych-out-dir /path/to/triptych/output \
  --intensity-channel 2 \
  --puncta-channel 1

This script:

• Determines whether each nucleus contains puncta
• Computes mean nuclear intensity
• Outputs a CSV containing:
  • Binary puncta label (present / absent)
  • Mean intensity
  • Source image metadata
  • Nucleus centroid coordinates
  • Optional triptych visualizations for quality control

---

4. C_sat estimation

In the same folder, open:

csat_estimation.ipynb

In the first cell, update:

CSV_PATH = "/path/to/output.csv"

Then run the notebook cell by cell.

This notebook:

• Fits a logistic curve
  • X: mean nuclear intensity (from nucleus channel)
  • Y: puncta presence (1 = present, 0 = absent)
• Produces diagnostic plots
• Estimates the intensity-based C_sat

---

Notes

• GPU usage is optional but recommended for Cellpose.
• Channel indices (--intensity-channel, --puncta-channel) should be adjusted based on OME-TIFF channel ordering.
• The nucleus pipeline represents the most up-to-date and validated workflow in this repository.