small-fish-gui 2.1.5

Small Fish is a python application for the analysis of smFish images. It provides a ready to use graphical interface to combine famous python packages for cell analysis without any need for coding.

Small Fish - A User-Friendly Graphical Interface for smFISH Image Quantification

Small Fish is a python application for smFish image analysis. It provides a ready to use graphical interface to synthetize state-of-the-art scientific packages into an automated workflow. Small Fish is designed to simplify images quantification and analysis for people without coding skills.

Cell segmentation is peformed in 2D and 3D throught cellpose 4.0+(published work) : https://github.com/MouseLand/cellpose; compatible with your own cellpose models.

Spot detection is performed via big-fish a python implementation of FishQuant (published work) : https://github.com/fish-quant/big-fish.

The workflow is fully explained in the wiki ! Make sure to check it out.

What can you do with small fish ?

✅ Single molecule quantification
✅ Transcriptomics
✅ Foci quantification
✅ Transcription sites quantification
✅ Nuclear signal quantification
✅ Signal to noise analysis
✅ Cell segmentation
✅ Multichannel colocalisation

Raw 3D fish signal with dapi

2D segmentation

3D Spot detection

Cluster detection

Analysis can be performed either fully interactively throught a Napari interface or performed automatically through a batch processing allowing for reproducible quantifications.

Installation

General setup

If you don't have a python installation yet I would recommend the miniconda distribution; but any distribution should work.

It is higly recommanded to create a specific conda or virtual environnement to install small fish.

As of version 2.1.0 Small Fish runs on python 3.12.

If you are using conda or miniconda

conda create -n small_fish python=3.12
conda activate small_fish

If you are using venv, after installing the official python 3.12, python-venv and python-tk .

python3.12 -m venv python_env/small_fish
source python_env/small_fish/bin/activate

Then download and install the small_fish package with :

pip install small_fish_gui

Setting up GPU

As of Small Fish 2.0.1 it is highly recommanded to set up GPU with cellpose since its new model, CellposeSAM, is very heavy computationally even more when attempting 3D segmentation.
First of all, try to run small fish gpu without additional commands depending on your configuration it could work straight out of the box. If encoutering any issue try first the following :

pip install --index-url https://download.pytorch.org/whl/cu124 torch torchvision torchaudio

If running into additional problems please look at cellpose documentation.

Run Small fish

First activate your python environnement :

conda activate small_fish

Then launch Small fish :

python -m small_fish_gui

You are all set! Try it yourself or check the get started section in the wiki.

Developement

Bugs to fix :

• Use of load button during co-localization quantification yields diffrent results than when testing from memory : --> Shown results are correct and consistent with results from memory --> But background (i.e cell_id = 0) is treated as a cell instead of being filtered --> But cells on edges are NOT removed from quantification (they are if test is performed from memory) --> But Cluster related metrics are not computed when test is performed from save