Table 1.
Structural coverage of the disease-related protein interaction networks analyzed in this work.
Table 2.
Detailed analysis of the location of nsSNPs in the disease-related protein interaction networks, based on complex structures and modelled interactions.
Fig 1.
Distribution of nsSNPs in the protein interaction networks of six selected diseases.
Distribution of nsSNPs (detailed for disease, polymorphism and unclassified) in the protein interaction networks from the six selected diseases, as classified in core, interface and surface non-interface, with expected distributions as calculated from residue composition, and odds ratios (OR) for the different residue locations and types of nsSNPs, based on (A) structural data; (B) modelled interactions; and (C) combined structural data and modelled interactions. Only significant OR values (P < 0.05) are shown.
Fig 2.
Prediction of interface residues and nsSNPs.
(A) Prediction success (sensitivity and precision) of interface residues using pyDockNIP (alone or extended with neighbor residues) on the proteins of the protein-protein docking benchmark 4.0, according to NIP cutoff value. (B) Interface and nsSNPs predictions using the extended pyDockNIP predictions on the proteins of the structural interaction networks from the six selected diseases. The nsSNPs predictions are detailed for interface disease-related, polymorphism and unclassified nsSNPs.
Fig 3.
Structurally unexplained pathological mutations of the RAS/MAPK pathway that are predicted to be involved at protein-protein interfaces.
Proteins of the RAS/MAPK pathway are represented as circles, showing pathological mutations that were not previously characterized due to the lack of structural data, but that have been predicted here to be binding hot-spots when docking with specific protein partners from this pathway (circles) or from the first-degree interaction network (in cyan squares). These docking partners thus represent proteins whose interaction is predicted to be affected by the mutation to which they are linked. Thus, all edges here correspond to interface predictions from docking.
Fig 4.
Pathways affected by pathological mutations in RAS/MAPK proteins predicted to be at binding hot-spots.
Proteins of the RAS/MAPK pathway are shown as colored circles, showing pathological mutations that were not previously characterized due to the lack of structural data, but that have been predicted here to be binding hot-spots for docking partner proteins involved in other pathways (linked to the corresponding mutation). Pathways shown in red are those that could not have been found using only available structural data.
Fig 5.
LHON interaction network with disease-related nsSNPs located at modeled protein-protein interfaces.
Proteins associated with LHON pathology (circles) and their modeled interactions (edges) with other proteins of the network (squares), showing the disease nsSNPs (for LHON and other pathologies) that are located at the modelled protein-protein interfaces. The homology-modeled structures for selected proteins are shown in ribbon, with disease nsSNPs in CPK representation, and all residues colored according to their NIP value (in red NIP > 0.2; in blue NIP < 0.0).
Fig 6.
Effect of nsSNPs in TNNC1 interaction network based on complex structures and modelled interactions.
Analysis of TNNC1 interaction network by combining structural data and docking models can identify different nsSNPs that could affect the interaction with different proteins. Protein-protein interactions with available structure are represented as red edges. Modelled interactions are represented as cyan edges. Selected protein-protein complex structures are shown, with residue color coding for the predicted NIP values as in Fig 5.