Integrative machine learning and gene regulatory network analysis identifies novel genes for leafy head formation in Brassica rapa and B. oleracea.

Abstract

Leafy head formation is a crucial developmental process in Brassica crops. Here, an integrative approach combining machine learning and gene regulatory network analysis was employed to identify novel genes involved in leafy head formation in Chinese cabbage (Brassica rapa) and cabbage (Brassica oleracea). Random Forest models, trained with 47 known leafy head-related genes, demonstrated robust performance with mean AUC values of 0.87 and 0.85 for B. rapa and B. oleracea, respectively. By filtering the model predictions, we identified 11 genes predicted with high confidence which were shared between both species. To further reveal the regulatory mechanisms, we constructed gene regulatory networks for genes in both species. By integrating ML predictions with these networks, we identified key regulatory clusters specifically related to leafy head formation. Network centrality analysis revealed many core genes in key clusters, including important transcription factors such as ANT, GRF2, GRF3, and TCX3, suggesting crucial roles in leafy head formation across the two species. The parallel detection of the same genes and similar network structures in the two species supports the validity of our findings. Our integrative approach provides novel insights into the genetic regulation of leafy head formation and sets the stage for future functional studies of Brassica species.