Thanks to advances in genetic sequencing, we can now identify proteins by their DNA. This has generated a tremendous amount of raw data to be explored. Biologists look at the interactions between proteins to generate protein-protein interaction networks. Even for the most-studied model organisms, these protein networks contain unlabeled or uncategorized vertices. A question of interest both biologically and mathematically is how to measure “similarity” between two vertices; a good metric would correctly categorize unlabeled vertices based on their distance to certain labeled vertices. Unlike in other naturally occuring networks (e.g., social networks), edges do not represent similarity but rather cooperation in some biological sense. Thus previously known clustering algorithms are ineffective. Here I will discuss one effective metric, diffusion state distance, as well as a new metric, exit frequency distance.