! ! Legume Genomics and Genetics 2025, Vol.16, No.1, 1-10 http://cropscipublisher.com/index.php/lgg! ! 7! Continuing Table1 SNP GWAS model (Ranking) LOD Allele Type Gene name Functional annotations Gm16_4935328 GLM, MLM, Blink(10), MLMM(22), SUPER(31), FarmCPU, CMLM(32) 5.61 T/G Glyma.16G051800, Glyma.16G052200 NAC domain protein; LRR- RLKs Gm19_44734953 GLM, MLM, Blink(3), FarmCPU, CMLM(4), MLMM(25) 6.02 G/A Glyma.19G189900, Glyma.19G190200, Glyma.19G190800 Defense response to bacterium; LRR- RLKs; plant-type cell wall Gm20_36724867 FarmCPU, CMLM(2) 6.54 C/T Glyma.20G124700 QSOX1 regulates plant immunity 6 Future Directions for GWAS in Soybean 6.1 Integrating multi-omics data If different types of data can be grouped together, such as the genome (DNA), transcriptome (RNA expression), proteome and metabolome (metabolites), the results of GWAS will be more convincing. This method can help us understand how complex traits come about from multiple perspectives and also narrow down some overly large associated regions. Sometimes, chain disequilibrium can make the range of results very wide. In the research of rapeseed, scientists used multi-omics analysis methods to overlap and compare QTLS found by different omics, successfully locating important genes related to nutritional metabolism and growth (Knoch et al., 2023). In soybean research, if a similar approach is also adopted, more useful candidate genes and signaling pathways may be discovered. In this way, the judgment of traits during breeding will be more accurate, and the efficiency of selection and breeding can also be improved. 6.2 Advancements in computational tools Nowadays, technology is becoming increasingly advanced, and there are already many new tools available for analyzing GWAS data. Especially some methods with machine learning (ML) capabilities perform very well when dealing with complex data. Previous GWAS methods did not perform well in analysis when there was an extremely large amount of data or when the genetic background of crops was relatively simple (such as soybeans). However, machine learning algorithms such as support vector regression (SVR) and random forest (RF) perform better in finding quantitative trait loci (QTL) (Yoosefzadeh-Najafabadi et al., 2021; 2023). These methods can also handle large volumes of data and make it easier to identify the genetic patterns behind complex traits. These tools also provide very practical technical support for research on genomic breeding. 6.3 Improving phenotyping techniques GWAS not only requires genetic data, but trait (phenotypic) data is equally important. Nowadays, an increasing number of studies are employing high-throughput phenotypic techniques to collect trait information from samples. Compared with traditional manual scoring, the new technology is more efficient and has less error. High-throughput phenotypes generally do not damage plants and can be continuously monitored, facilitating the observation of changes in traits over time (Xiao et al., 2021). For instance, scientists have employed deep learning (DL) technology to identify soybean disease conditions, with an accuracy rate even higher than that of manual judgment (Rairdin et al., 2022). If these advanced phenotypic methods are combined with GWAS, not only can the accuracy of the analysis results be improved, but also the traits with good performance can be selected more quickly, which is particularly helpful for the breeding of new soybean varieties. 7 Concluding Remarks Nowadays, when scientists study the genetic mechanism of soybeans, they often use a method called GWAS (Genome-wide Association Study). This method has been of great help. Through GWAS, researchers have identified many quantitative trait loci (QTLS) and also discovered some candidate genes related to yield, quality, plant height and disease resistance. For instance, one study utilized soybean varieties from Canada and China. They used GWAS to identify five gene regions related to yield, protein content and oil content. These findings are very helpful for soybean breeding. Another study combined the results of 73 GWAS for a pooled analysis, which
RkJQdWJsaXNoZXIy MjQ4ODYzNA==