Protein Interaction Network Screensaver

STRING is a database of known and predicted protein-protein associations, maintained by the STRING Consortium, a collaboration of the Swiss Institute of Bioinformatics, the Novo Nordisk Foundation Center for Protein Research and EMBL. Its website lists thousands of organisms, though this saver follows human genes.

Each link between two proteins carries a combined confidence score between 0 and 1 (in practice shown from 0.150 to 0.999), calibrated to approximate the probability that the association is real. It combines separately scored evidence channels: genomic-context predictions (gene neighbourhood, gene fusion, co-occurrence across species), laboratory experiments, curated pathway databases, automated text mining and co-expression. The thickness of an edge here shows that score. An association means two proteins are functionally linked, such as sharing a pathway, and not necessarily touching physically.

The layout is a force-directed graph: connected nodes pull toward each other like springs while all nodes repel, so tightly linked groups cluster together. The position on screen carries no biological meaning beyond that clustering. A sparse neighbourhood can simply reflect a gene that has been studied less, not one that does little.