Bianca Stöcker presents her work “Protein Complex Similarity Based on Weisfeiler-Lehman Labeling” at the “Similarity Search and Applications” (SISAP) conference at Newark, New Jersey, USA.

Protein complexes consist of several protein chains. The similarity between two complexes can be measured by comparing the sets or multisets of their protein chains, but it is conceivable that two complexes consist of the same (multi)set of chains, but has a different topology, i.e. different physical interactions between the chains. This should be reflected in the similarity measure. Our work defines such a similarity measure that incorporates topological information but is almost as efficiently computable than set-based measures.

This publication resulted from a collaboration within the DFG-funded Collaborative Research Center SFB 876 between members of projects A6 (Resource-efficient graph mining) and C1 (Feature selection in high dimensional data for risk prognosis in oncology). This work is also part of our long-running project on protein hypernetworks and constraints between protein interactions.