Protein-protein interaction networks data

The two protein-protein interaction networks used in this work were constructed using the following human datasets: PINA and STRING [48, 49]. STRING describes 263,666 interactions between 14,732 proteins from the integration of: BIND, DIP, GRID, HPRD, IntAct, MINT and PID databases [50–55]. PINA includes 108,477 unique interactions between 15,450 different proteins collected from six publicly available and manually curated databases: IntAct, MINT, BioGRID, DIP, HPRD and MIPS/MPact [56]. Only direct physical interactions were used in this study, avoiding both data derived from phylogenetic studies (preventing tautologies in the results when comparing with tree distances) and interactions obtained by textmining processes [57].

PINA covers 63% of the proteins present in the Ras phylogenetic tree and 31% of all possible connections between them, while STRING covers 77% and 52% respectively. Although the PPI data from PINA and STRING integrate a similar source of information (physical interactions, as mentioned), they show a different level of coverage of the Ras tree data and also a different network topology. Therefore, both were considered as valid and complementary datasets in this analysis.