The invention relates to a prediction method for the ubiquitination degradation of a target protein. The prediction method comprises the following steps: A1, predicting the interaction of protein subunits by adopting a convolutional neural network method; A2, constructing a spatial three-dimensional structure of the ubiquitin marking system; A3, predicting a ubiquitination-degraded target protein,downloading a three-dimensional structure file of the target protein, complementing the three-dimensional structure of the target protein by using a homologous modeling method, and predicting the interaction between the target protein and a E3 ubiquitination marking system by using the convolutional neural network method; after that, optimizing an overall spatial three-dimensional structure of interaction between the E3 ubiquitin marking system and the target protein by adopting a molecular dynamics method, judging whether the target protein is possibly marked by the E3 ubiquitin marking system or not according to the stability of the system, and performing ubiquitin-target protein marking by virtue of the ubiquitin-target protein marking system, and degrading the protein by virtue of a ubiquitin-proteasome.