Aggregation / Solubility Redesign
This method searches for hydrophobic patches on the surface of a protein of interest. Then Rosetta Design tools selectively substitute amino acids in patches with charged residues, making mutations to preserve overall stability while optionally limiting the extent of mutation away from the native sequence. This method has been experimentally validated and shown to improve both the solubility and stability of proteins.
Lau et al (2018) JBC 293 p. 13224 and Chennamsetty et al (2010) J Phys Chem B 114 p. 6614
Immunogenicity Prediction predicts likely MHC II epitopes using a machine learning algorithm trained on known MHC II epitopes published in King et al, PNAS 111 p. 8577 (2014).
Reactive Residue Prediction & Removal
Reactive Residue Prediction & Removal uses standard bioinformatics approaches from sequence to identify likely sites for: proteolysis, deamidation, glycosylation and oxidation. By using these as a filter on a design output, or by designing a region to remove propensity for any of these characteristics (e.g. to design out a proteolysis site), a user can remove these identified chemical modification regions. Identification can be on the basis of sequence rules combined with solvent accessibility, although more elaborate predictors have started to appear
See, e.g. Chennamsetty et al (2015) J Pharm Sci 104 p. 1246 and Sydow et al (2014) PLoS One 9 e100736.