| Title: | Graph Kernels for Biological Network Analysis |
| Speaker: |
Karsten Borgwardt
Computational and Biological Learning Lab, The University of Cambridge |
| Abstract |
A wide variety of different algorithms have been developed to
study cellular networks such as protein-protein interaction networks.
These algorithms focus on finding motifs, modules and pathways within
these networks, or on aligning them. In bioinformatics, however, little
work has been done in the field of global network comparison which
measures overall similarity of two networks.
Graph kernels, a recent trend in machine learning, allow us to measure global similarity between biological networks in a principled manner. They not only provide an overall similarity score for two networks, but as part of the family of kernel machines, they enable us to perform tasks such as classification, clustering, feature selection and two-sample tests on networks. Hence graph kernels might become an important tool in biological network analysis and systems biology over coming years. In this talk, we will present our work on developing graph kernels, and outline how graph kernels can contribute to the field of biological network analysis.
|