Predicting Molecules: Zou and MDockPP's Journey

CAPRI challenges are a way for teams to prove their skills in predicting how biomolecules will work together without any prior knowledge. This story looks at two teams, humans lead by Zou and a server group run by MDockPP, as they took on 9 rounds of these CAPRI challenges known as 47-55. One of the key changes in the competition came when AlphaFold2, a powerful tool for predicting protein structures with high accuracy, was released to the public. Before this, these teams did not have a lot of luck predicting accurate models for how biomolecules fit together. They used a method called homology modeling and docking which was not very effective It was not until AlphaFold2 was able to be used by all that it changed how the teams predicted biomolecule formations. What did it do? Zou's human team saw a boost in accurate predictions, going from 1 correct interface prediction out of 19 to 4 out of 8. The improvement was undeniable. This approach helped a lot with specific targets, like T231 and T233, providing medium-quality models that weren't possible with the old methods. What caused the improvement? The ability to try many samples at once, or massive sampling, was a big help. It allowed for better, more detailed predictions. The study shows that new tools and techniques can make a big difference and that it is important to use multiple approaches to achieve the best results

The science of biology has gone through innovation after innovation, much like this story's take on computer modelling. Just think of this progression. Imagine a scientist long ago discovering a new organism or chemical substance. There was very little they could do with the stuff. They could not predict any important reactions or interactions from the new discovery. The problem was more knowledge. How can we predict what the discovery will do? That is all the progress of biology. We try to use all the information we get in a way to make accurate predictions about what is going to happen when we change a variable. We have moved from just having clean tables to the ability to view the changes more visually on screen instead of written out. We have improved the tools so much, that we have predictive abilities! Where will it all go next? Question: What if we took all that AlphaFold2 provides us and think about the next steps? Maybe we could figure out how biomolecules work together even better, and use that to create new medicines or treatments. Even better, maybe we could use this information to save the world from a big problem and we would do it much quicker than before.

Actions