Spirov Alexander V, Holloway David M
State University of New York at Stony Brook, Computer Science Department and Center of Excellence in Wireless & Information Technology, Stony Brook University Research & Development Park, 1500 Stony Brook Road, Stony Brook, NY 11794-6040, USA.
Procedia Comput Sci. 2010 May 1;1(1):999-1008. doi: 10.1016/j.procs.2010.04.111.
A new approach to design a dynamic model of genes with multiple autonomous regulatory modules by evolutionary computations is proposed. The approach is based on Genetic Algorithms (GA), with new crossover operators especially designed for these purposes. The new operators use local homology between parental strings to preserve building blocks found by the algorithm. The approach exploits the subbasin-portal architecture of the fitness functions suitable for this kind of evolutionary modeling. This architecture is significant for Royal Road class fitness functions. Two real-life Systems Biology problems with such fitness functions are implemented here: evolution of the bacterial promoter rrnPl and of the enhancer of the Drosophila even-skipped gene. The effectiveness of the approach compared to standard GA is demonstrated on several benchmark and real-life tasks.
提出了一种通过进化计算设计具有多个自主调控模块的基因动态模型的新方法。该方法基于遗传算法（GA），并特别为此目的设计了新的交叉算子。新算子利用亲本字符串之间的局部同源性来保留算法找到的构建块。该方法利用了适用于此类进化建模的适应度函数的子盆地-门户架构。这种架构对于皇家道路类适应度函数具有重要意义。这里实现了两个具有此类适应度函数的实际系统生物学问题：细菌启动子rrnPl的进化和果蝇even-skipped基因增强子的进化。在几个基准和实际任务中证明了该方法与标准GA相比的有效性。
