Summary
High-pLDDT designs, particularly those generated by hallucination, can be insoluble (1,2). (3) found that inversion of AlphaFold2 led to many surface-exposed hydrophobic residues, and that manual post-design intervention led to in vitro expression of some designs. Alternatively, inverse folding can circumvent this (see Inverse folding methods outperform sequence design by hallucination).
1.
Dauparas J, Anishchenko I, Bennett N, Bai H, Ragotte RJ, Milles LF, et al. Robust deep learning–based protein sequence design using ProteinMPNN. Science. 2022;378(6615):49–56. Available from: https://doi.org/10.1126/science.add2187
2.
Wang J, Lisanza S, Juergens D, Tischer D, Watson JL, Castro KM, et al. Scaffolding protein functional sites using deep learning. Science. 2022;377(6604):387–94. Available from: https://doi.org/10.1126/science.abn2100
3.
Goverde CA, Wolf B, Khakzad H, Rosset S, Correia BE. De novo protein design by inversion of the AlphaFold structure prediction network. Protein Science. 2023;32(6). Available from: https://doi.org/10.1002/pro.4653