Outlook on Future Problems in RNA Design

Evolution in vitro can be subdivided into two subclasses of processes that are analogs of natural and artificial selection: (1) RNA evolution in replication assays without intervention and (2) optimization according to predefined external selection criteria in cycles as shown in Fig. 2.1. The latter case can be viewed as "breeding of molecules." In biotechnology evolutionary design stands in competition with rational design as done, for example, by inverse folding. The advantage of the evolutionary strategy consists in the fact that only the desired function has to be known and testable, whereas no information on structure is required. The success of rational design, on the other hand, depends critically on the state of our knowledge of the relations between structure and function. Admittedly, this knowledge is fragmentary at present but it is rapidly increasing.

Function predicted from known structure has the opposite problem: many structures may give rise to the same function. If one day we are in a position to derive structures required for given functions with fair reliability, then the rational design problem will be solvable since we could then invert our initial two-step relation:

sequence v structure v function

If the inverse problems were solved, rational design would be superior to its evolutionary counterpart because the effort needed to design structures for functions and to synthesize sequences that are expected to have the required structures is minimal compared with the effort of screening 1015 different molecules or more. For the time being, however, the evolutionary method in most cases is the only one likely to produce the desired outcome.

As in other scientific disciplines, inverse methods are coming increasingly into the focus of interest and new algorithms are being developed that provide tools for reverse engineering. The inverse folding algorithm described here is presumably the simplest example of such a strategy in the RNA world. Other reverse engineering problems, for example the design of sequences for given spectra of suboptimal conformations or the creation of molecules that show predefined kinetic folding behavior, are much more ambitious and harder to solve. Needless to say, to conceive and implement an algorithm for the universal design of RNA structure for RNA function seems to be a far-fetched problem. Applied mathematicians, on the other hand, have lots of experience in the development of inverse methods and the tasks to be solved invite truly interdisciplinary approaches.

0 0

Post a comment