Skip to main content
. 2023 Sep 7;6:188. doi: 10.1038/s42004-023-00991-6

Fig. 7. The overall workflow for the MULTICOM protein tertiary structure prediction system.

Fig. 7

The single-chain structure prediction process consists of five sequential steps: (1) multiple sequence alignment sampling, (2) template identification, (3) monomer structural model generation, (4) structural model ranking, and (5) Foldseek-based iterative model refinement. If the target is a chain of a protein assembly, the target and other chains in the protein assembly are also fed into the assembly structure prediction module built on top of AlphaFold-Multimer to generate quaternary structures. The tertiary structures of the target are then extracted from the predicted quaternary structures and are added to the structural model pool generated by the single-chain protein structure prediction.