In order to determine the accuracy of the functional predictions of proteins found through the augmented network, protein degrees were calculated in order to compare both the data in the original interactome and in the bacterial meta-interactome. Each protein was labeled with the most probable function and then compared to the number of degrees it contained. The number of degrees a protein has equates to the number of interaction partners or relationships it may have.

Fig. 6: Calculation of Correctly Predicted Functions. The prediction of proteins with a low degree was improved with the addition of the meta-interactome.

The provided heat map portrays the different functions that a given protein can have. The s-index is provided as well and each blue marker on the heat map represents a protein. A random sampling was done regarding the pool of proteins that were found to be poorly characterized, or believed to still have unknown functions. This included 342 proteins of the entire data set. A Z test was applied to this random sampling 100 times for efficiency. Predictions were concluded to be found in 299 of those 342 proteins.

Fig. 7: Scores of functional proteins