Machine Learning for Protein Engineering
The age of machine learning is overtaking us all rapidly, both increasing our capabilities while undercutting what we thought we were doing as scientists. The Ellington lab's journey in this regard began with the development of a convolutional neural net, MutCompute, by some very talented graduate students. We have applied and continue to apply MutCompute to a variety of protein engineering projects. In parallel, MutCompute has undergone its own evolution, eventually resulting in an algorithm for the prediction of protein stability, Stability Oracle. Advances waiting in the wings include the use of machine learning methods to traipse through chemical / ligand space, and opportunities on the horizon for hallucinating new enzymes.
Hosted by Enver Izgu
~Coffee/tea will be served prior to the lecture~