Application of Markov Stability for graph-based clustering on protein-protein interaction networksTools von Holy, Peter (2023) Application of Markov Stability for graph-based clustering on protein-protein interaction networks. MRes thesis, University of Nottingham.
AbstractProtein-Protein interaction networks are one of the most well-explored and documented parts of the interactome, as such, they have had a variety of databases and analyses developed for them, in order to harness this highly useful abstraction of very complex systems. Community detection is a popular analysis for many datasets which can be abstracted onto graphs and otherwise is a concept still performed on non-graph-based datasets through clustering methods. Community detection can also be performed at varying scales through the introduction of artificial time parameters, which in this case is a result of the use of a measure called Markov Stability. Markov Stability is also used as a measure to define a good graph partition but optimizing by having it be the objective function of the Louvain algorithm. In this study, we implement a framework for multiscale community detection governed by Markov stability, which has been previously used in other studies and apply this framework to an example protein-protein network of the proteins related to the 20 most frequently mutated human cancer genes from the STRING database. The results of this application are then explored and we show that due to the underlying properties of the example, robust partitions are obtained across varying Markov times.
Actions (Archive Staff Only)
|
Tools
Tools
