Biasing Scoring Weights

[scottyler89]

Sorry if this already exists, but:

To my eye in 540, if others are getting similar results to me, it seems as though too much weight is given to intramolecular bonds, which don't contributing to the affinity between the two proteins. I think it would be a good beneficial for these "design the interface" types of puzzles to "bonus points" or have a multiplication factor for bonds, or especially hydrophobic interactions, which contribute to intermolecular binding.

The Janeway Immuno book also says that antibody-antigen binding surfaces, more than ordinary protein - protein interactions, have a high propensity for aromatic residues.

For the antibody, or small inhibitory peptide design puzzles, another good biasing measure may be to increase the weight given to intermolecular "packing" and "hiding" scores.

~Scott

[beta_helix]

assigning…

[Aaron Chevalier]

That's a good idea and we have discussed up weighting the interface residues. However, with last helix puzzle the results were so promising(especially the best designs) we decided against constraining this puzzle . Specifically, the best designs from the last helix puzzle showed a lot of aromatics(Tyr/Trp/Phe) at the interface so we decided that it was best just to leave the puzzle as is.

Also, you are right on about antibodies having lots of aromatics and its something we keep in mind on our end when we pick which designs to test. We just like seeing lots of diverse solutions at the starting rounds of design in case the FoldIt players find something we miss.