I am working on computational protein design and protein aggregation is a big problem that I face. If there is a way to predict whether or a protein will aggregate (or its probability) my work will become more efficient.
I don't believe that there is a direct way to predict protein aggregation. If the protein is folded and exhibits a high surface charge it is very unlikely. In the unfolded state practically all proteins are becoming susceptible to aggregation. Generally, the number of water exposed hydrophobic groups should be a key determinant.
There are different forms of aggregation - one is the native protein exceeding its solubility - here, total charge and surface hydrophobicity play a role - a protein is least soluble close to its isoelectric point, and a high surface charge can keep protein in solution.
In protein design, you face a different types of aggregation problem:
- misfolding of your protein either leading to exposure of hydrophobic residues that ought to be buried in the core,
- "domain swapping", where secondary structure elements that ought to interact with the rest of the molecule in an intramolecular fashion do so in an intermolecular fashion,
- and, as a third possibility, in particular sequences with a high beta sheet propensity forming amyloid -like extended intermolecular beta sheets. For the third, there are some prediction algorithms around, e.g. Article Prediction of Peptide and Protein Propensity for Amyloid Formation
Article Detecting hidden sequence propensity for amyloid fibril formation
https://omictools.com/protein-aggregation-prediction-category
But keep in mind that prediction algorithms are just that, predictions. They are a measure of what we know, and, in their many failures, the many things we don't know about protein folding.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Probability