GIBBER

• Genes
• In
• Brain
• BrowsER

About

Exploitation of disorder - gene - brain structure relations, determined by phenotype studies and spatial transcriptomic, is a recent, but quickly advancing topic in the medical sciences. The exploration of these datasets is, however, currently limited to only a few pre-selected diseases or manually selected genes with, with no provenance information, let alone visual links. We therefore propose a novel system integrating these aspects to browse genes and disorders in the brain, GIBBER. The system provides expert a top-down selection process to move from disorder hierarchies to comparing brain structures' activations by genes based on two genome association studies. This flow is enabled by, first selecting full disorder hierarchies, yielding gene selections from genome association studies, and compare them, followed by a configurable result view to investigate the gene expression by structure, both spatially in slice- and volumetric views, and in dimensionality-reduced representations. The system was developed through an integrated process with experts, re-evaluating our requirements as needed, to arrive at a useful and effective tool for their analysis.

(Abstract from our forthcoming paper about GIBBER)

You can already check out a live version hosted at the HEREDITARY Demo Server!

Getting started

Installing the Backend dependencies and getting the data

Prerequisites: uv and docker

Once you have these installed you can go ahead and start the development Docker containers, and download all dependencies for the backend:

(cd data/dev-containers && docker compose up -d)
uv sync

Then you can start downloading the necessary data for the ingestion:

source ./.venv/bin/activate
cd backend/preparation
python3 00_load_all.py

Please note that you have to download the Snomed CT hierarchy yourself from https://download.nlm.nih.gov/umls/kss/IHTSDO2026/IHTSDO20260601/SnomedCT_InternationalRF2_PRODUCTION_20260601T120000Z.zip, as they are licensed, and extract them to data/snomed_ct. Once the data files are ready, you can finally start the (lengthy) ingestion process.

python3 01_ingest.py
python3 02_gwas.py
python3 02a_snomed.py

Once that is done you can start the backend using

cd ..
python3 -m uvicorn backend:app --reload --port 8093 

Setting up the frontend

Prerequisites: some recent LTS node version

cd frontend
npm i -D
npm run dev

TODOs

• CSV Download
• Scrollable MRI Slices
• GWAS data integraion
• PCA attributions back into genes view