MBKGVISUAL

Version 1.0.0

By Skye Lane Goetz

MicrobiomeKG Visualization Tool. MBKGVisual reads compressed graph data (NODES and EDGES TSV files), builds a NetworkX graph, samples it using degree-based neighbor expansion (0.1% of nodes), and generates an interactive HTML visualization using Pyvis.

Live Demo

GitHub Pages Deployment

Usage

Install Dependencies

pip install pyvis networkx polars pandas pyarrow backports.zstd scipy

Run Visualization

python3 ./main.py

This reads ./NODES.tsv.zst and ./EDGES.tsv.zst and outputs ./index.html.

Architecture

The visualization pipeline is implemented in main.py:

1. read_tsv - Decompress zstd TSV files into Polars DataFrames
2. is_sig - Filter edges where significant == "YES"
3. sanitize - Remove invalid nodes ("1", "2", "3", "4") and self-loops
4. is_directed - Split edges: biolink:affects is directed, others are undirected
5. mkgraph - Build NetworkX DiGraph (undirected edges get symmetric pairs)
6. sampler - Degree-based neighbor expansion sampling (0.1% of nodes)
7. mkvis - Generate interactive HTML via pyvis

Contributors

Skye Lane Goetz