Whole interactome predictions.

For each organism investigated, we aggregated the data we used for training and testing, trained a final model and predicted the whole interactome of that organism. Organism: latin name for eight model organisms sorted alphabetically; Nprot: number of proteins in proteome (values taken from [28]); NpredPPI: subset of PPIs used for prediction in which both proteins are dissimilar to the proteins in the positive interactions of the training set; NprotPred: corresponding number of proteins for which NpredPPI interactions were predicted, see Eq 1 for calculation; NpredPPI novel: denotes the number of predicted PPIs for which both proteins are dissimilar to any known positive interaction, including redundant and low-quality PPIs; NpredPPI ProfPPIdb: subset with strongest predictions of previous column contained in our resource; Nprot ProfPPIdb: number of unique proteins in the PPIs published at https://rostlab.org/services/ppipair/, as well as https://figshare.com/collections/ProfPPI-DB/4141784.