However, it can still serve as a rough guide for the performance benefits of a fully dynamic scheme over static clustering and that can be used for comparison with our less complex rotating clustering method.
We have reported in detail evidence for the numerous well-studied complexes detected by clustering methods.
It is essential to eliminate the noise proteins and the interactions from the complexes created via clustering methods.
So in Table 5, we present the number of matched clusters and matched complexes predicted by the clustering methods on the other six PPI networks.
The outcome demonstrates the benefits of non-stationary feature fusion with clustering methods for complex data analysis where existing approaches do not exhibit a high performance.
We have shown that our approach, which uses non-binary co-complex information, is superior to clustering methods that dissect binary protein protein interaction data.
For the complexes which are predicted by clustering methods, only a small portion of them can be formed in a continuous time point set, and the average overlapping rates during the formation are significant lower than that of the known complexes.
Although the existence of the protein complexes has not yet been validated experimentally, the compatibility between the prediction of protein complexes by our clustering method and the consistency of GO terms offers reliable candidates for novel functional protein complexes to be validated by future experiments.
Markov Cluster Algorithm (MCL) [ 12, 13] has been one of the most successful clustering methods in identifying complexes from protein interaction networks.
The complexes of each clustering method before and after refinement are denoted as algorithm_O and algorithm_R.
For convenience sake, the complexes predicted by each clustering method are mentioned as original complexes, the complexes refined by our method are referred as refined complexes, denoted as algorithm_O and algorithm_R in tables and figures, respectively.
