A Tour of the Human Protein-Protein Interaction Network

The Problem.

The human protein-protein interaction network is incredibly vast and information dense: STRING contains over 19,000 proteins, collectively comprising millions of interactions. The objective of this project was to create an immersive experience of the human PPI network that gives a sense of its sheer scale while providing a view into its underlying structure.

Building a hierarchical network.

The entire human ppi network is far too large to visualize all at once. To organize the network, we leveraged a "community detection" method: An algorithm that identifies highly connected communities within a complex network. Each community is connected to the others and holds within it a rich network of protein-level interactions.

Diving inside.

Using atomistic-level data from AlphaFold, we created a protein-level tour of each community, capturing protein-level interactions and the visual identity of each protein.

Identifying Function.

Within every community, the protein-protein interaction network is further organized into sub-communities. Using bioinformatics approaches (gene ontology analysis), we highlighted the functional groupings that emerged from the network subcommunities.