基于双层粒子群优化算法的弥漫大B细胞淋巴瘤基因表达样本分类研究

Classification of Diffuse Large B Cell Lymphoma Gene Expression Data Samples Based on The Two-Layer Particle Swarm Optimization

目的从分子生物学的角度对弥漫大B细胞淋巴瘤(DLBCL)基因表达样本进行分类,改进基本粒子群优化分类算法在分类精确性和稳定性方面的不足。方法给出一种基于双层粒子群优化(TLPSO)的分类算法,选取141个弥漫大B细胞淋巴瘤样本,包括氧化磷酸化(OxPhos)、B细胞受体(BCR)和宿主反应(HR)3种亚型,随机选取训练集和测试集以获取不同样本组合,与基本粒子群优化(PSO)分类算法进行比较。结果基于TLPSO的分类算法获得较好分类结果,最佳分类预测结果数和分类结果分布2项指标均优于PSO算法。结论双层粒子群优化分类算法能够对弥漫大B细胞淋巴瘤基因表达样本进行准确和稳定分类,能为临床肿瘤基因表达样本的分类定型提供依据。

Objective To determine subtype of diffuse large B cell lymphoma （DLBCL） gene expression sam- ples from molecular biology level, and to improve the particle swarm optimization （PSO） classification algo- rithm in both accuracy and stability. Methods A classification algorithm based on the two layer particle swarm optimization （TLPSO） was proposed. 141 DLBCL samples, which included oxidative phosphorylation （OxO- hos）, B-cell receptor/proliferation （BCR） and host response （HR） subtype, were selected randomly to com- pose uncertain training and test sample subgroups. The PSO classification algorithm was used for comparison. Results The TLPSO based classification algorithm outperforms PSO in two indicators, which were the best pre- diction result and the distribution of classification results. Conclusion The TLPSO based classification algo- rithm could provide precise and stable classification results for tumor gene expression samples. It could serve as a basis for determining the subtype of clinical tumor samples.