Table 1: Modalities of STRING data used for protein function prediction.

For each modality of STRING, the description and number of features are included in the table.

Modality	# features	Description
PPI	17,901	Protein-protein interactions
Curated database	10,501	Interactions in curated databases
Co-expression	18,618	Co-expression relationships between proteins
Neighborhood	3,974	Co-location in the same genomic neighborhood
Co-occurrence	2,936	Co-occurrence of orthologs in other genomes
Fusion	6,053	Fusion of orthologs in other genomes