P3 - Predicting Portable Promoters

Development of bioinformatic tools to design DNA sequences that can be used control elements to optimize the production of desired compounds in industrial microorganisms.

In recent years, DNA synthesis technology has become drastically cheaper. This allows us to engineer life at increasingly large scales, combining genes found in various organisms to construct new biological functions in microorganisms such as bacteria or yeast. Remarkable breakthroughs of such synthetic biology have been reported. However, a key remaining challenge is that while we know how to write the DNA, we do not know exactly what to write. Attempts at designing biological circuits often still take many man-years of development or even fail completely, because it is not well understood how DNA sequences of genes found in other organisms should be adapted and combined. There is a clear need for rational design methods for synthetic biology, allowing carefully constructed DNA building blocks to be combined for optimal, robust and predictable functionality in a variety of organisms. In this project, we will address one of the main needs in synthetic biology applications: to obtain predictable and robust gene expression at desired levels in industrially relevant microorganisms. To this end, we will develop a new class of bioinformatics algorithms to design full DNA sequences for promoters, the regions that control the expression of genes. The key novelty of our approach is (a) to exploit machine-learning at the core of these algorithms, rather than heuristics or approximate biological models; and (b) to use yeast, the model eukaryotic microorganism, as a "skunk works" for constructing promoters for less easily accessible microbes. We will generate a specific set of thousands of promoters tested for activity in yeast to learn to predict expression given a DNA sequence; once this predictor is available, we can then use optimization techniques to design DNA sequences yielding desired relative levels. These designs will be validated, both in yeast and in the industrially highly relevant fungus Aspergillus niger.