用Boltzmann系综质心预测RNA-RNA相互作用结构
Prediction of RNA-RNA interaction structure by centroids in the Boltzmann ensemble
摘要：
新的高通量测序技术使得全基因组转录组学的出现成为可能。这一进展与最近发现的调节性非编码RNA（ncRNA）相结合，需要快速准确的算法来预测RNA-RNA相互作用的概率和结构。虽然有算法来预测两个核酸的最小自由能相互作用二级结构，但很少有人利用碱基对概率中的信息来改进相互作用结构预测。在本文中，我们提出了一个算法来预测的汉明质心的玻尔兹曼系综的相互作用结构。我们还提出了一个有效的算法来采样的相互作用结构的合奏。我们的采样算法使用一个平衡的计划遍历索引，提高了运行时间的丁劳伦斯采样算法。Ding-Lawrence采样算法具有$O(n^2m^2)$时间复杂度，而我们的算法具有$O((n+m)^2\log(n+m))$时间复杂度，其中$n$和$m$是输入链的长度。我们在新版本的{ \tt piRNA}中实现了我们的算法，并将我们的结构预测结果与竞争对手进行了比较。我们的质心预测平均优于竞争对手的最小自由能预测算法。
英文摘要：
New high-throughput sequencing technologies have made it possible to pursue the advent of genome-wide transcriptomics. That progress combined with the recent discovery of regulatory non-coding RNAs (ncRNAs) has necessitated fast and accurate algorithms to predict RNA-RNA interaction probability and structure. Although there are algorithms to predict minimum free energy interaction secondary structure for two nucleic acids, little work has been done to exploit the information invested in the base pair probabilities to improve interaction structure prediction. In this paper, we present an algorithm to predict the Hamming centroid of the Boltzmann ensemble of interaction structures. We also present an efficient algorithm to sample interaction structures from the ensemble. Our sampling algorithm uses a balanced scheme for traversing indices which improves the running time of the Ding-Lawrence sampling algorithm. The Ding-Lawrence sampling algorithm has $O(n^2m^2)$ time complexity whereas our algorithm has $O((n+m)^2\log(n+m))$ time complexity, in which $n$ and $m$ are the lengths of input strands. We implemented our algorithm in a new version of {\tt piRNA} and compared our structure prediction results with competitors. Our centroid prediction outperforms competitor minimum-free-energy prediction algorithms on average.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
