The statistics of directed evolution: From library generation to high throughput screens

Directed evolution has been recognized as a powerful approach to creating enzymes and cells with desirable properties, such as growth on a new substrate, thermotolerance, or even novel reaction mechanisms. The directed evolution approach is to generate diversity through genetic mutations and screen the diversity for mutants that have improved properties compared to the parent. This diversification and screening can be repeated to generate mutants with superior properties. While the concept is straightforward, the devil is in the details. How should I create diversity (error prone PCR, mutagenic primers, transposons…)? How much diversity do I need (103, 106, 107, more??). If my high throughput screen (HTS) does not find an improved mutant, does that mean an improved mutant did not exist in the library? Or could it mean my HTS failed to isolate it? How do I identify single cells with 20% improvement, when cell-to-cell variability is more than 10 fold larger? In this workshop, we will start to answer several of these questions. We will review approaches for generating and quantifying diversity. We will develop a Bayesian framework for understanding noise due to cell-to-cell variability and be able to predict the expected enrichment achievable in fluorescence-activated cell sorting (FACS) and other HTS screens. We will discuss common pitfalls in generating libraries and screening them. If you do directed evolution on enzymes or cells, if you use FACS, droplet-based HTS, or even plate-based screening, this workshop will provide you with fundamental tools to add rigor to your directed evolution efforts and improve your likelihood of success