LIVIA — Local Interaction VIsualization and Analysis

Analyze protein-protein interactions from structure predictions and generate visualization scripts

Prediction Analysis FlyPredictome Ortholog Interactome AFDB Dimer AFDB Monomer Subdomain cLIP PPI Network Tutorials About GitHub
📄 Kim & Perrimon (2026). LIVIA: a browser-based tool for assessing and visualizing predicted protein interactions. bioRxiv. →

Prediction Analysis

Analyze predictions from AlphaFold-Multimer, AlphaFold3, ColabFold, Boltz-1/2, Chai-1, OpenFold3, Protenix-v2, and ESMFold2. Calculate interface confidence metrics (iLIS, LIS, cLIS), view interactive PAE/LIS/cLIS maps, and generate visualization scripts.

Launch Prediction Analysis →
Supports any major structure prediction platform — no installation needed.

cLIP beta (clustered Local Interaction Profiler)

Cluster a protein's interactome by shared contact Local Interaction Residues (cLIR) — drop lis.py output, all in your browser.

Launch cLIP →
Resolves AlphaFold DB structures automatically from sequence or protein name.

PPI Network Builder beta

Build an interaction network from lis.py output — Leiden communities run in your browser.

Launch PPI Network →
Interactive layouts, name→symbol auto-convert, and export to Cytoscape/SVG.
Browse Precomputed Interactomes — FlyPredictome · AlphaFold DB

FlyPredictome

Search Drosophila protein interactions from FlyPredictome predictions (Kim et al., 2026).

e.g., Akt1–foxo, Dl–N, Dsh–Axn

Ortholog Interactome

Search human kinase–TF interactions from ortholog predictions (Kim et al., 2025).

e.g., TP53, CDK2, AKT1, BRAF

AFDB Dimer

Fetch dimer predictions from the AlphaFold Protein Structure Database (Han et al., 2026).

e.g., Fiber protein 2, pp71

AFDB Monomer Subdomain

Analyze intramolecular domain interactions in AlphaFold monomer predictions.

e.g., Haspin (Q8TF76), SCN9A (Q15858)

Tutorials

Visual walkthroughs for each tool — learn by example.

About

Metric definitions, color schemes, and references.

GitHub

Source code, documentation, and issue tracker.
Browser: Tested on Chrome and Safari (macOS/iOS). All analysis runs locally in your browser — no data leaves your device.
Note: Each prediction platform (AlphaFold3, ColabFold, Boltz, Chai-1, etc.) may produce different confidence calibrations. The iLIS ≥ 0.223 threshold was established using ColabFold/AlphaFold-Multimer predictions. Other platforms may require adjusted thresholds.
LIVIA GitHub · AFM-LIS GitHub
Kim & Perrimon (2026)
Ah-Ram Kim · Harvard Medical School