The increase of available Protein-Protein Interaction (PPI) data enables us to develop computational methods for protein complex prediction. A protein complex is a group of two or more proteins formed by stable interaction over time. In general, it corresponds to a cluster in a PPI Network (PPIN). However, clusters correspond to not only stable protein complexes, but also to sets of proteins with transient interactions. As a result, conventional graphtheoretic clustering methods that usually treat stable and transient interactions in the same manner show high false-positive rates in protein complex prediction. We propose an approach for refining a PPIN using structural interface data of protein pairs for protein complex prediction. The construction of the Simultaneous Protein Interaction Network (SPIN)is an essential step in our approach, which excludes competition between mutually exclusive interactions formed by an overlapping interface. We then use naive clustering algorithms on SPIN for the prediction of protein complexes. When we compared the prediction results of our method with a simple PPIN-based method, it turned out that the SPIN-based approach outperforms the simple PPINbased method in the sense that it predicts all of the complexes the conventional method generates, and even more true-positive ones.
Use Patent Claims
Include Install Instructions
These details are provided for information only. No information here is legal advice and should not be used as such.