基于多解并行遗传算法预测RNA二级结构及假结

Prediction of RNA Secondary Structure and Pseudoknot based on Multisolution Parallel Genetic Algorithm

非编码RNA功能通常与其结构密切相关,准确预测RNA的二级结构有助于揭示RNA的功能。在传统的遗传算法基础上,结合RNA二级结构和假结的特点,提出一种多解并行的遗传算法预测RNA二级结构和假结。首先构建茎区池和初始解集;然后根据RNA二级结构的基本特点建立目标函数、约束函数、自适应度函数和相应的遗传算子,基于初始解集中多条序列—结构模型进行并行遗传迭代,预测最优RNA二级结构;然后在RNA二级结构基础上,使用遗传算法继续进行迭代和筛选,预测含假结的RNA二级结构。实验结果表明,该方法不仅可以解决大规模茎区的组合问题,还可以减少随机性。该方法与常用的预测假结的IPknot方法比较,对单序列RNA的结构预测结果正确率高且稳定。

Non-coding RNA function is usually closely related to its structure,and accurate prediction of RNA secondary structure can help reveal the function of RNA.In this paper,a multi-solution parallel genetic algorithm for predicting RNA secondary structure and pseudoknot is proposed based on the traditional genetic algorithm,combining the characteristics of RNA secondary structure and pseudoknot.Firstly,the stem region pool and initial solution set are constructed;then the objective function,constraint function,adaptive degree function and corresponding genetic operators are established according to the characteristics of RNA secondary structure,and the optimal RNA secondary structure is predicted by parallel genetic iteration based on multiple sequence-structure models in the initial solution set;Based on the RNA secondary structure,the genetic algorithm is then used to continue iteration and screening to predict the RNA secondary structure containing pseudoknot.The experimental results show that this method can not only solve the combination problem of large-scale stem regions,but also reduce the randomness.compared with the IPknot method,this method is more acourate and stable in predicting the structure of single sequece RNA.