Collaboration With McGill
We collaborated with McGill University’s iGEM team, who also shared the same focus of using computational tools to engineer proteins. Their project involved engineering a set of proteins to break down cholesterol, in which they started with enzymes that catalyze the transformation of testosterone and evolved the protein to break down cholesterol. Their computational algorithm involved using the FITTED algorithm (Molecular Forecaster) co-developed by Nicolas Moitessier at McGill University.
Our collaboration consisted of cross validation to computationally predict how well each team’s mutations achieved the goal on the other team’s pipeline. On Harvard’s end, we assessed the mutations produced by McGill’s team by determining the binding energies through the Prodigy software we used in our pipeline. However, instead of computing the binding of a receptor to a peptide-ligand (as was done with our own project), we expanded our approach to compute the binding of the protein to a non-peptide ligand, which served as an indicator of it being favorable for testosterone to fit within the active site of McGill’s engineered proteins. The results of this cross validation are shown below:
McGill’s team explained that this data, “helped us prioritize which mutants to test on the GCMS so that going forward we'll test the mutants with higher affinity for our substrates." On McGill’s end, they used their FITTED algorithm to structurally and energetically test if our peptide ligand docked in the correct location that we had predicted using AlphaFold. Their pipeline applied to our project enabled a structural verification to classify how well our AlphaFold structures predicted the binding location of the ligand on the receptor, which we were unable to capture in our own computational approach (which focused more on binding energy).
We would like to thank McGill for collaborating with us to advance both of our projects.
Image: Left (Members from McGill University), Right (Members from Harvard University)