Principles of <i>in vitro</i> selection of ribozymes from random sequence libraries

<i>In vitro</i> selection methods are used to identify catalytic RNAs from pools of random sequences. We discuss the central concepts using experimental data and computational models. Experiments proceed in multiple rounds, each with a reaction step and a step in which reacted sequences are recovered. Sequences are enriched each round by a factor depending on combined reaction and recovery probability. In the first round, there are few functional sequences, and it is necessary to minimize the probability of losing these. In later rounds, the loss probability is negligible, and the procedure can be optimized to maximize the enrichment factor. Clusters of related sequences emerge which descend from separate sequences in the initial pool. The fitness of an RNA depends on how well it matches a structure with specified sequence and base-pair constraints. Sequences that exactly match the constraints may be rare, but sequences a few mutations away are much more common; hence it is likely that clusters descend from suboptimal sequences. There is a high probability that beneficial mutations arise during the experiment. This explains the experimental observation that there is little correlation between cluster frequencies and fitnesses, whereas correlation between enrichment factors and fitnesses is strong.