The application discloses a method for predicting different metabolic states of cells in
cancer and metabolic characteristic subtypes of the cells based on a
metabolite-
protein interaction network, and the method comprises the following steps: (1) identifying the classification and
metabolic state of the cells: preprocessing data, filtering low-quality cells; performing twice
principal component analysis according to
key genes; performing twice dimension reduction clustering on the cells respectively; (2) predicting metabolic characteristic subtypes of a
queue according to the proportion of different metabolic states of cells in the sample. The method can identify the
metabolic state of the cells and metabolic subtypes, and provides a new perspective for
cancer research.