An integrated WGCNA-XGBoost framework for prioritizing candidate stress-responsive hub genes in soybean under nutrient and heavy metal stress conditions.

Abiotic challenges such as nutrient shortages and heavy metal toxicity dramatically limit soybean yields worldwide. While many transcriptomic investigations have examined responses to single stresses, few have identified regulatory nodes that function across multiple adverse conditions. In this study, we conducted a multi-dataset transcriptomic meta-analysis integrating six RNA-seq experiments comprising 66 samples across nutrient deficiency and heavy-metal stress conditions. Following batch-effect correction using ComBat and variance-based filtering, weighted gene co-expression network analysis identified stress-associated modules. XGBoost was then applied within these modules to prioritize genes with high predictive importance. Prioritized candidate gene hub genes were defined using the combined criteria of strong module membership (|kME| > 0.8) and high XGBoost feature importance. Comparative synteny analysis across leguminous species further supported the evolutionary conservation of several prioritized candidates. These genes represent high-priority targets for future functional validation and crop-improvement studies aimed at enhancing soybean stress resilience.Model robustness was assessed using 5-fold cross-validation and receiver operating characteristic (ROC) analysis, with several genes achieving area under the curve (AUC) values exceeding 0.90. Gene set enrichment analyses revealed coordinated regulation of hormone signaling pathways, cell wall remodeling, nitrate transport, and MAPK signaling cascades. Further synteny analyses showed that several of the identified hub regions are conserved in related legume species, including Phaseolus vulgaris, Vigna unguiculata, and Vigna angularis, highlighting the evolutionary conservation of these stress-responsive modules. Overall, the integrated WGCNA-XGBoost framework identifies conserved regulatory hubs involved in multi-stress responses and provides a powerful systems-biology strategy for prioritizing candidate genes and guiding the development of stress-tolerant soybean cultivars.