ABSTRACT
Emerging RNA-based approaches to disease detection and gene therapy require RNA sequences that fold into specific base-pairing patterns, but computational algorithms generally remain inadequate for these secondary structure design tasks. The Eterna project has crowdsourced RNA design to human video game players in the form of puzzles that reach extraordinary difficulty. Here, we present an eternamoves-large repository consisting of 1.8 million of player moves on 12 of the most-played Eterna puzzles as well as an eternamoves-select repository of 30,477 moves from the top 72 players on a select set of more advanced puzzles. On eternamoves-select, a multilayer convolutional neural network (CNN) EternaBrain achieves test accuracies of 51% and 34% in base prediction and location prediction, respectively, suggesting that top players’ moves are partially stereotyped. We then show that while this CNN’s move predictions are not enough to solve numerous new puzzles, inclusion of six additional strategies compiled by human players solves 61 out of 100 independent puzzles in the Eterna100 benchmark. This EternaBrain-SAP performance is better than previously published methods and in the middle of the performance range of newer algorithms developed by Eterna participants and other groups. Our study provides useful lessons for efforts to achieve human-competitive performance with automated RNA design algorithms.