Large-scale studies of protein complexes often involve affinity purifications of tagged-proteins ‘one-at-time’ followed by LC-MS/MS. Recently, large-scale coverage of the complexome could be achieved through protein correlation profiling (PCP), without the need for tagged-proteins. A typical PCP experiment involves the use of size exclusion chromatography to separate protein complexes based on their size. Proteins in the same complex are co-eluted and are likely to have high correlation in protein abundance profiles across all the fractions. The aim of this study is to compare three different correlation metrics for the identification of protein complexes in Saccharomyces cerevisiae. These metrics include the Pearson correlation coefficient, Spearman's rank correlation coefficient, and Maximal Information Coefficient (Reshef et al. 2011). A high confidence set of protein complexes in S. cerevisiae curated by Benschop et al. (2010) was used for benchmarking. From the analysis of seven PCP datasets, the Spearman’s correlation consistently identified more protein complexes with higher average correlation per complex, outperforming the other two metrics. An application of PCP is to identify changes in protein complexes between mutant and wild type yeast strains, which could be identified through changes in Spearman’s correlation pattern between mutant and wild type. Two knockout mutants of lysine protein methyltransferases (efm4∆ and efm7∆), which solely targets the methylation of eukaryotic translation elongation factor 1α (eEF1α), affected the correlation profile of proteins in the eEF1α complex. The knockout of arginine protein methyltransferase (hmt1∆), which catalyze the methylation of Npl3p, also led to changes in Npl3p’s correlation with known partners. The above suggests protein methylation could affect protein-protein interactions and complex formations. Future directions would involve peaks identification from protein abundance profiles to reduce noise (Scott et al. 2014) and the use of machine learning to predict direct protein-protein interactions (Drew et al. 2017).