IEEE/ACM Trans Comput Biol Bioinform. 2023 May-Jun;20(3):2170-2176. doi: 10.1109/TCBB.2022.3233844. Epub 2023 Jun 5.
Various diseases severely affect maize, leading to a significant reduction in yield and crop quality. Therefore, the identification of genes responsible for tolerance to biotic stress is important in maize breeding programs. In the present study, a meta-analysis on microarray gene expression of maize imposed to various biotic stresses, induced by fungal pathogens or pests, was performed to identify key tolerant genes. Correlation-based Feature Selection (CFS) was performed to attain fewer DEGs discriminating control and stress conditions. As a result, 44 genes were selected and their performance was confirmed in the Bayes Net, MLP, SMO, KStar, Hoeffding Tree, and Random Forest models. Bayes Net outperformed the other algorithms representing an accuracy level of 97.1831%. Pathogen recognition genes, decision tree models, co-expression analysis, and functional enrichment were implemented on these selected genes. A robust co-expression was observed among 11 genes responsible for defense response, diterpene phytoalexin biosynthetic process, and diterpenoid biosynthetic process in terms of biological process. This study could provide new information on the genes responsible for resistance to biotic stress in maize to be implicated in biology or maize breeding.
各种疾病严重影响玉米，导致产量和作物质量显著下降。因此，在玉米育种计划中，鉴定对生物胁迫具有耐受性的基因是很重要的。本研究对玉米受到真菌病原体或害虫等各种生物胁迫的基因表达进行了微阵列元分析，以鉴定关键的耐受基因。采用基于相关性的特征选择（CFS）来获得更少的差异表达基因（DEGs），以区分对照和胁迫条件。结果，选择了 44 个基因，并在贝叶斯网络、MLP、SMO、KStar、Hoeffding 树和随机森林模型中验证了它们的性能。贝叶斯网络优于其他算法，代表的准确性水平为 97.1831%。对这些选定的基因进行了病原体识别基因、决策树模型、共表达分析和功能富集。在防御反应、二萜类植物抗毒素生物合成过程和二萜类生物合成过程方面，有 11 个负责的基因之间观察到了稳健的共表达，这是生物过程的共表达。本研究可以为玉米对生物胁迫抗性的基因提供新的信息，这些信息可能涉及生物学或玉米育种。
