Maize Genomics and Genetics 2025, Vol.16, No.6, 316-324 http://cropscipublisher.com/index.php/mgg 319 why co-expression network analysis is receiving increasing attention in stress research. Through these networks, the gene modules that "advance and retreat simultaneously" during high-temperature stress and filaments development can be extracted and then connected with phenotypic data, which basically can narrow down the candidate range. Many studies have proved that such networks can locate some regulatory centers. For instance, certain core transcription factors are often found at the "crossroads" of these networks, and they have a significant impact on flowering time and adaptability to adverse conditions (Schaefer et al., 2017; Zhou and Liang, 2024). Once these networks are linked to eQTL or phenotypic QTL data, it becomes more confident to identify those genes and regulatory relationships that play key roles. 4.3 Predictive models based on genomic selection (GS) and machine learning When dealing with complex traits, traditional models often "neglect one aspect for another". But now, genomic selection (GS) and machine learning (ML) are gradually becoming powerful tools to address this challenge. They can not only model with whole-genome marker data, but also introduce transcriptome and epigenome data to improve prediction accuracy. Especially for traits like silk spinning time that are influenced by multiple factors, they show obvious nonlinearity at high temperatures. At this time, deep learning or random forests can come in play and perform more stably than linear models. However, these models are not a universal template, provided that the quality of the omics data is good enough. Overall, once multi-omics information can be smoothly embedded into GS and ML systems, it will be a considerable boost for the rapid screening of high-potential genotypes and the improvement of heat-resistant breeding efficiency. 5 Functional Validation and Molecular Mechanism of Key Candidate Genes 5.1 Screening and annotation of candidate genes within QTL regions Identifying candidate genes that affect silk spinning under high-temperature stress, relying solely on QTL mapping is clearly insufficient. Usually, methods such as GWAS, linkage mapping and meta-QTL need to be used in combination in order to screen out those truly potentially useful genes from the huge data, especially those that are actively expressed during the flowering stage (Longmei et al., 2021; Djaloviac et al., 2023). Members like ZmBAG-8 and ZmBAG-11 in the BAG family, as well as ZmLACS9 in the LACS family, have drawn attention due to their active performance under high-temperature conditions. They are not only significantly upregulated under stress, but have also been found to be involved in processes such as protein folding, lipid metabolism and ROS clearance. The annotation results also suggest that they possess typical stress response regulatory motifs. However, not all candidate genes have the same high "appearance rate" as them, and the screening process always involves many trade-offs. 5.2 Functional validation via expression profiling and transgenic approaches Whether the candidate genes are "really working" or not needs to be proved by experiments. qPCR and RNA-seq are conventional methods for initial analysis, but to really get to the bottom of it, one still has to resort to genetically modified organisms. Take ZmBAG-8 and ZmBAG-11 for example. The expression levels of these two significantly increased after stress in heat-resistant strains and their F1, but decreased when in heat-sensitive materials, indicating that they are indeed related to heat resistance (Farid et al., 2025). The situation of ZmLACS9 was similar. When mutant experiments were conducted, it was found that it was prone to damage chloroplasts and increased ROS accumulation under heat stress, and its performance was worse than that of the wild type (Wang et al., 2023). Some protein-protein interaction network analyses have even pushed these genes to the "central position" of regulation. When these results are combined, it can basically be said that these genes are involved in regulating the heat resistance response during silk spinning. 5.3 Regulatory network construction and integration with signaling pathways The matter is not as simple as "a certain gene is upregulated under stress". Many candidate genes are actually located in a very complex signaling pathway. To understand their roles, we need to take the regulatory network out and look at it together. For instance, tools such as co-expression analysis and eQTL networks have revealed that some genes with silk priority expression are actually remotely regulated by trans-factors, not just by their own "rotation". Interestingly, genes such as ZmHsftf13 or Bx10 are not only related to heat stress, but also involved in
RkJQdWJsaXNoZXIy MjQ4ODYzNA==