exomewalker logo

Prioritization of whole-exome data by random-walk analysis of protein-protein interactions


Exome Walker combines the filtering and prioritization of variants in an exome sequence with gene based prioritization by random walk analysis (RWA). RWA exploits the protein protein interaction (PPI) network to characterize the similarity between genes (and the proteins they encode). Intuitively, RWA defines an adjacency matrix based on PPI data (we use PPI data from the STRING database), and then uses matrix methods to explore all paths between all pairs of genes. Two genes that are highly interconnected, say by a direct interaction and multiple paths of two or three interactions, but are not well connected to other proteins will be assigned a higher degree of similarity that proteins that are not well connected to one another or that have interactions with large numbers of other proteins.

Our methodology was initially described in Köhler et al. (2008). In that work, we used RWA to prioritize genes located in a linkage interval. Here, we combine RWA with standard filtering procedures for exome data based on rarity, variant quality, predicted pathogenicity, and inheritance pattern. See Smedley, Köer et al. (2014) for further details (manuscript submitted).