Network Intelligence

Computational biology • network science

Discover the proteins
hidden inside disease networks.

Explore how protein interactions and computational prediction methods can help identify candidate proteins for further research.

This is a research and educational computational tool, not a clinical diagnostic system. Rankings are candidate hypotheses, not diagnoses or treatments.

Interactive computational reproduction of Agrawal, Zitnik & Leskovec, Large-scale analysis of disease pathways in the human interactome (PSB 2018), using the SNAP disease-pathway PPI and DisGeNET association datasets. Documented method deviations are listed on the Methodology page.

Measured sample · Alzheimer pathway

This is a real subset of existing SNAP edges. Decorative drawings elsewhere on this page are not data.

01

Disease

Choose a disease

Start from a named condition in the research catalogue.

02

Protein Network

Explore how proteins interact

See the protein–protein interaction map for that disease pathway.

03

Prediction

See which proteins are ranked highly

Inspect candidate proteins ranked by several computational methods.

04

Graphlets

Explore small patterns in the network

Compare disease-level orbit profiles — the small building blocks of a network.

How it works

From a disease name to a ranked candidate list

Think of a disease as a starting list of proteins. Those proteins sit in a larger map of protein partnerships. The computer studies that map, ranks other proteins that may be relevant, and then looks at small repeating patterns.

  1. 01Disease
  2. 02Protein interactions
  3. 03Disease network
  4. 04Prediction methods
  5. 05Combined ranking
  6. 06Network patterns
  7. 07Candidate proteins

This strip is a teaching diagram of the project workflow. It is not a new scientific result.

A simple story

What this website is showing

The drawings below are educational illustrations. They are not measured protein interactions and they are not prediction results.

Think of proteins as tiny machines inside cells.

In this website every protein is labeled only by its Entrez Gene ID.

Proteins interact with each other.

A line means a recorded protein–protein interaction from the SNAP interactome, not a new lab experiment.

A disease can be associated with a group of proteins.

Those associated proteins form a disease pathway: a small subgraph of the larger map.

Our models rank other proteins that may be worth investigating.

A high rank is a computational hypothesis — a candidate, not a cause or a treatment.

Graphlets describe small patterns inside the network.

They are educational building blocks. Disease-level 73-orbit profiles are not the same as the 12-d protein motif proxy.

Protein–protein interaction (PPI)

Lines in the measured maps are recorded protein–protein interactions from the SNAP / BioGRID interactome, not new lab experiments run on this website.

Candidate protein

A high rank means the computational model placed that protein near the top. It does not mean the protein causes or cures the disease.

Combined ranking

Five methods each give a score. The Combined Score is the mean of those scores after each method is stretched onto a 0–1 scale.