A disproportionate number of predicted proteins from the genome sequence of the protozoan parasite Trypanosoma brucei, an important human and animal pathogen, are hypothetical proteins of unknown function. This work describes a protein correlation profiling mass spectrometry approach, using two size exclusion and one ion exchange chromatography systems, to derive sets of predicted protein complexes in this organism by hierarchical clustering and machine learning methods. We provide examples of both potential new subunits of known protein complexes and of novel trypanosome complexes of suggested function, contributing to improving the functional annotation of the trypanosome proteome. These hypothesis-generating proteomic data are provided in an open access online data visualisation environment (http://134.36.66.166:8083/complex_explorer)