Abstract: Artificial intelligence accurately predicted the three-dimensional structure of proteins for the first time, which was selected as one of the top ten scientific breakthroughs in 2020 by "Science" magazine, and became a frontier direction in the field of structural bioinformatics. Most single-chain proteins in nature contain multiple domains. In a biological sense, inter-domain association and cooperation are crucial to achieve multiple related functions. This paper firstly introduces the development of protein structure prediction and the critical assessment of structure prediction (CASP); Secondly, some representative methods are briefly described in three parts: single-domain protein structure prediction methods, multi-domain protein structure assembly methods and end-to-end protein structure prediction methods; The commonly used databases and model quality evaluation indicators in protein structure prediction are then demonstrated, and the performances of the representative prediction methods are compared. Finally, we conclude with a brief overview of the future challenges and outstanding questions in the field.