BioSET Visualizer

Introduction

Project is being actively developed.

Setup

Clone (including the Biomni submodule)

git clone --recurse-submodules https://github.com/Chahat08/BioSET_Visualizer.git

If you already cloned without --recurse-submodules, initialise the submodule afterwards:

git submodule update --init --recursive

To pull the latest changes including any submodule updates:

git pull
git submodule update --recursive --remote

BioSET Python environment

python -m venv .venv
# Windows: .venv\Scripts\activate
source .venv/bin/activate

pip install -U pip
pip install -e .

Running

The rendering is performed at the server's side. The following starts a local server instance, and opens the client in a browser (at http://localhost:8080/index.html).

source .venv\bin\activate
# Windows: .venv\Scripts\activate
bioset

The default port is 8080. To change this, you can supply a port command line argument such as bioset --port 1234.

By default, console output is suppressed. To enable logging/print statements, use the --logs flag:

bioset --logs

To run the server remotely, and prevent the browser from opening up, do:

source .venv\bin\activate
# Windows: .venv\Scripts\activate
bioset --server --host 0.0.0.0 --port 1234

Then, on the client machine, navigate to http://<server-ip-address>:1234/index.html in a browser.
If you are not able to access the server ensure that the server is reacheable from the client network and that there is no firewall blocking access.

AI Assistant (Biomni)

BioSET includes an AI assistant powered by Biomni for biological labelling and free-form queries about active markers.

Prerequisites

• Conda (Anaconda or Miniconda)
• Anthropic API key

Environment setup

Biomni runs in its own conda environment, separate from the main BioSET environment.

cd Biomni/biomni_env
conda env create -f environment.yml
conda activate biomni_e1

API key

Copy .env.example to .env and fill in your Anthropic API key:

cp .env.example .env
# then edit .env and set ANTHROPIC_API_KEY=sk-ant-...

Starting the Biomni server

The Biomni server must be running before using the AI assistant in BioSET. Start it in a separate terminal with the biomni_e1 environment active:

conda activate biomni_e1
cd BioSET_Visualizer   # repo root
python Biomni/run_server.py --port 5000

The server listens on http://localhost:5000 by default. Leave this terminal open while using BioSET.

Using the AI assistant

1. Start BioSET normally in a separate terminal (bioset --logs)
2. Load data and activate channels
3. Open the AI Assistant panel in the right drawer
4. Click Initialize — this connects to the local Biomni server and loads the agent (takes mutiple minutes on the first run while datasets download)
5. Use Label to generate biological labels for the currently active markers (screenshot is sent automatically)
6. Type a free-form question and press Send to query the agent about the active markers

Config

Most relevant settings such as link to the zarr, channel indices and voxel spacing can be changed in app.py main() function.

Refer to config.py for a full list of possible settings and their meanings.