Studies have given tantalizing hints that natural selection can distinguish between enzyme variants (Benner 1989; Dean 1989). But such studies have been necessarily limited to either a few easily obtainable natural variants or to a smattering of mutants created by site-directed mutagenesis. To detect underlying trends and determine whether this is in fact generally true, the scope of such studies will need to be widened to examine a broad spectrum of random variants (Dykhuizen and Dean 1990). The development of in vitro evolution techniques has given scientists the powerful tools necessary to undertake such a study. Vast numbers of mutants can be quickly generated in the lab, selected for function, assayed to determine catalytic characteristics, and placed into competition experiments against wildtype cells to determine relative fitness. For the first time, it is possible to empirically determine the relationship between enzyme function and fitness.
