cellColor 0.2.0

An app for exploring spatial transcriptomics and cell segmentations by overlaying data

Gene Visualization Tool (cellColor)

An interactive desktop application for visualizing spatial transcriptomics data, cell segmentation masks, cell clustering data and microscopy images. Perfect for verifying cell segmentation accuracy and exploring spatial gene expression patterns.

---

👥 Credits & Project Team

• Developer: Anthea Guo
• Mentor: Kushal Nimkar
• Principal Investigator (PI): Prof. Karthik Shekhar

---

Motivation

The motivation behind this was to create a tool that allows the user to easily verify the accuracy of cellpose segmentation masks while also highlighting spatially unusual genes.

---

✨ Features

• **Image Loading & Zooming:**Load tissue/microscopy images, zoom into regions, and reset to full view.
• **Cellpose Segmentation Overlay:**Overlay Cellpose-generated segmentation masks or outlines with smooth, cached zooming.
• **Transcript Visualization:**Import transcript coordinates (x, y, gene), align with images using transformation matrices, and overlay selected genes.
• **Single-Cell Integration:**Load AnnData cell center positions (.h5ad), toggle display, and customize appearance.
• **User-Friendly Toolbar:**Intuitive controls for overlays and zoom, live status feedback, and collapsible navigation frames.
• Data Alignment: Load transformation matrices for accurate transcript-image alignment.

---

🚀 Installation

Option 1: Local Development (Editable Mode)

git clone https://github.com/crocodile27/cellColor.git
cd cellColor
conda create -n cellcolor python=3.10
conda activate cellcolor
pip install -e .

Run locally:

cellColor

Option 2: Install via PyPI (v0.1.0)

pip install cellColor

Release: September 1, 2025 (PyPI link)

Launch:

cellColor

---

‼️ IMPORTANT: Required Data Formats

Images

Tissue Section Images: .png, .jpg, .tif. High resolution tissues that will act as the canvas and which images you choose is up to your discretion. Click on file-> load image and choose desired image. A downsized version of the image will be produced after it is loaded for the first time. Feel free to choose either of them in future runs.

All other data:

Place all of the following files in a single folder with the format **[prefix]***rn[insert run number]*_**rg[insert region number]**. E.g. 140g_rn3_rg0 -> _this is really important for the auto_load_file function to correctly detect with run and region you are working on and automatically load all necessary files.

Files include:

• Cellpose Masks: .npy image mask generated from cellpose.
• Detected Transcripts: CSV/TSV with barcode_id, global_x, global_z, x, y, fov, gene, transcript_id as columns. global_x & global_y will be used as the coordinates.
• Transformation Matrix: CSV/TSV for alignment of gene and tissue data.
• AnnData:.h5ad with cell center coordinates saved in global_x and global_y observations.

---

🧪 Example Workflow

1. Open the app: cellColor
2. Load image: File → Load Image to display tissue section.
3. Auto load all files: → click on desired folder
4. Overlay Data: choose the data you'd like to overlay including: cell centers, gene transcripts, cellpose masks/outlines, cell clusters.
5. Zoom & reset: Zoom into areas of interest; use Reset Zoom to return to previous zoom level.