The approach used for these submitted sequences was motif scaffolding with RFdiffusion of the binding region/motif loop of EB2 (the natural receptor/binder) extracted from the 2VSM conformation. The generated backbones/designs were of lengths between 60 and 200 residues. Then, proteinMPNN was used to generate the sequence of the whole protein (including redesigning the EB2 motif) at 3 different sampling temperatures (0.1, 0.3, 0.45) with 100 generated sequences per sampling temperature. Then, the sequences with a better proteinMPNN score than the median value were selected. Finally, sequences were clusterized at different sequence identities depending on the protein length of the backbone from RFDiffusion to have more than 15-20 residue differences between them.
No proteins found matching your criteria.