Hofacker Ivo L
Department of Theoretical Chemistry, University of Vienna, Vienna, Austria.
Methods Mol Biol. 2014;1097:71-84. doi: 10.1007/978-1-62703-709-9_4.
In this chapter we present the classic dynamic programming algorithms for RNA structure prediction by energy minimization, as well as variations of this approach that allow to compute suboptimal foldings, or even the partition function over all possible secondary structures. The latter are essential in order to deal with the inaccuracy of minimum free energy (MFE) structure prediction, and can be used, for example, to derive reliability measures that assign a confidence value to all or part of a predicted structure. In addition, we discuss recently proposed alternatives to the MFE criterion such as the use of maximum expected accuracy (MEA) or centroid structures. The dynamic programming algorithms implicitly assume that the RNA molecule is in thermodynamic equilibrium. However, especially for long RNAs, this need not be the case. In the last section we therefore discuss approaches for predicting RNA folding kinetics and co-transcriptional folding.
在本章中，我们将介绍通过能量最小化进行RNA结构预测的经典动态规划算法，以及该方法的变体，这些变体允许计算次优折叠，甚至是所有可能二级结构上的配分函数。后者对于处理最小自由能（MFE）结构预测的不准确性至关重要，并且例如可用于推导为预测结构的全部或部分赋予置信值的可靠性度量。此外，我们还将讨论最近提出的替代MFE标准的方法，例如使用最大预期准确度（MEA）或质心结构。动态规划算法隐含地假设RNA分子处于热力学平衡状态。然而，特别是对于长RNA，情况未必如此。因此，在最后一节中，我们将讨论预测RNA折叠动力学和共转录折叠的方法。
