Pierre Baldi has been applying Deep Learning, a type of AI, to study scientific problems, like protein folding. How is AI impacting science?
AI is being used on differential gene expression data sets to start answering a whole host of interesting questions. [Don’t read beyond this point if you are triggered by misuses of the term -omics]
Okay, it seems that the AlphaGo algorithm has reached 90 percent success at predicting how a protein will fold. But what would happens if we ask AlphaGo to process random aa sequences? IOW, could AlphaGo helps in assessing how rare protein folds are in sequence space?
Another interesting question is how it’s ability to predict folding is affected by different folding conditions, such as temperature, pH, and water activity.
My overly pessimistic and completely unfounded opinion is no. I would suspect that many functional protein folds in random sequence are going to be transient, sensitive to environment (as @Rumraket mentions), and variable even within a single random sequence. In addition, there are cofactors (e.g. ATP, divalent cations) and chaperone proteins to consider. My experience with expressing recombinant proteins has taught me that it isn’t as simple as having the right amino acid sequence.