Mistakes in genome design
By Joel Bader
Computational design of DNA sequences with defined function is a goal of synthetic biology, but we still lack crucial information required to construct objective functions reflecting reality. Thus, much work in synthetic biology still relies on synthesize -and-screen rather than design-and-build. Because of known unknowns and unknown unknowns, design challenges at the genome level require tradeoffs between the benefits of more ambitious designs and the risks of fatal flaws. The complete synthesis of the yeast genome by the international Saccharomyces cerevisiae (Sc2.0) consortium has reached a milestone of 3.5 MB out of 12 MB finished in the form of entire synthetic chromosomes replacing their wild-type cognates, which in turn provides a first opportunity to characterize design flaws. We provide an overview of our mistakes, including both systematic errors and random failures, and discuss how we would have revised the design had we known then what we know now. Ageneral conclusion is that our bug rate was very low, about 5x10^-5 in fitness defects per bp changedin protein-coding regions, and that our design was much more robust than we had anticipated.