The Simpson's-index was calculated for the S. pneumoniae proteins in the network in order to determine the heterogeneity of their function probabilities. For example, for a given protein if its Simpson’s s-index was to be near 1, that means there was a function that dominated the other potential functions in that given probability set. If the s-index was closer to 0, then it can be assumed that there is a rough equal distribution of the probability of functions that could be identified with that given protein.

The orange identifies merely the S. pneumoniae network whereas the green identifies the S. pneumoniae network plus the meta-interactome that helped identify functions for unknown proteins. With this comparison to the right, it can be seen that the s-index is much higher when the meta-interactome is taken account of. The x-axis identifies the degree as the number of interactions a protein will have. For example, a degree of 3 means that a protein has three different interactions with three other proteins in that given network.

Fig. 7: S-index of predicted functions

The mean s-indices were calculated for each protein, as shown in figure D to the right. This figure demonstrates that the accuracy of a dominant function existing for every unknown proteins was strengthened when the data of the meta-interactome was added to the S. pneumoniae protein network.

Fig. 8: Comparison of s-indices

When adding the meta-interactome data to the original interactome, the question about accuracy arises, if the information about the functions of the unknown proteins are assumed correctly. Figure E to the right helps reflect the answer to that question and ultimately helps illustrate how close the function “answers” are with just the original network and with the added meta-interactome.

Both axes list the different functions that the proteins may have and the p-value identifies the predicted functions for each given interactome. It is found that the majority of the functions predicted in the original network corresponded to the functions in the augmented network, therefore we can believe that the accuracy of the augmented network is relatively high.

Fig. 9: Determination of Function The functions of each previously-unknown protein was determined using the meta-interactome and augmented network.