基于花授粉算法的蛋白质网络功能模块检测方法

揭示未知蛋白质功能是后基因时代蛋白质组学中的核心内容之一,运用群集智能思想识别蛋白质相互作用网络(protein-protein interaction network, PPIN)中的功能模块已经成为该领域的一个研究热点。提出一种基于花授粉算法(flower pollination algorithm, FPA)的蛋白质相互作用网络功能模块检测方法(FPA for functional module detection in PPIN, FPA-FMD)。采用随机游走的方式对种群中的每个花粉进行编码,并利用花授粉算法特有的自花授粉和异花授粉机制优化种群,其中自花授粉采用重组策略和取优策略,异花授粉采用基于Levy机制的变异策略和基于差异度的自适应变异策略,4种策略分别从不同角度推进了种群的进化。在3个公共数据集上的仿真试验表明:与其他6种经典算法相比,本研究提出的算法的整体性能优良而且在F度量和准确度两项综合指标上具有绝对优势。

Flower pollination algorithm-based functional module detection in protein-protein interaction networks

Revealing unknown functions of proteins were one of the core contents of proteomics in the post gene era, where it had become a hotspot to use the swarm intelligence-based approaches to identify functional modules in protein-protein interaction networks(PPIN). An approach based on flower pollination algorithm to detect functional modules in PPIN was proposed. Each pollen in the population was encoded by a random walk and the population was optimized by using two mechanisms of self-pollination and cross-pollination which were specially owned by flower pollination algorithm. More specially, the strategies of recombination and better-solution selection were adopted in the self-pollination while the mutation strategies based on Levy mechanism and an adaptive individual-difference were employed in the cross-pollination. The four strategies together promoted the evolution of the population from different angles. The simulation experiments on three public data sets showed that the proposed algorithm had not only excellent overall performance but also absolute superiority in terms of two comprehensive indicators F-measure and accuracy compared with the other six classical algorithms.