Protein Design by Multi-Objective Optimisation
As presented in PyData London Meetup #58 (MAN, meetup.com)
Abstract:
When optimising for multiple factors, a common practice is to reduce the problem to 1D,
which, because of the limited search space, is likely to result in a suboptimal solution.
In Multi-Objective Optimisation we consider a set of trade-off solutions, called the Pareto Front,
as the optimal set, and by so explore a multi-dimensional space, where we are more likely
to find improved solutions.
Motivated by the challenge to empirically navigate the vast DNA combinatorial
sequence space to discover therapeutic proteins,
I discuss Pareto Front optimisation using Evolutionary Algorithms.
In particular, I highlight DEAP
, a Python platform for prototyping.
Watch on YouTube, or here: