This document discusses using classificatory decomposition (CD) to classify laser-induced breakdown spectroscopy (LIBS) data from four proteins into their respective types. CD uses multi-objective evolutionary algorithms and rough sets to decompose spectra into additive components to both reconstruct the data and test classification accuracy using few components. The authors hypothesize CD will effectively classify LIBS data of bovine serum albumin, osteopontin, leptin, and insulin-like growth factor II proteins, detection of which could aid in diseases like ovarian cancer. They evaluate accuracy using leave-one-out cross validation, involving training on all instances except one and testing on the left out instance.