Using Network Models Of Proteins To Predict Functionally Linked Interfaces Of Proteins (Flips) At The Residue Level

Proteins perform numerous cellular activities either individually or by interacting with other proteins. The function carried out by participating oligomers is representative of their structural stability and interaction specificity. One general goal of our research is to identify the properties of interacting amino acid residues that guide formation of Functionally Linked Interfaces of Proteins (FLIPs) upon oligomerization. In this work, proteins are represented as networks (Residue Interaction Networks, RINs) where residues are the nodes and residue proximity (both bonded and non-bonded interactions) define the edges. Prior studies of RINs by others have shown that network properties aid identification of active sites and residues important for protein folding. In our prior work, we could distinguish known FLIPs from Functionally unCorrelated structures (FunCs) using RINs with ∼73% accuracy and found FLIPs were generally much more organized than FunCs. In this work, we expand this concept to identify the threshold values of network features for individual residues participating in a FLIP interface such that potential FLIP interface regions can be predicted at the residue level.