Machine learning uncovers accumulation mechanism of flavonoid compounds in Polygonatum cyrtonema Hua

Abstract

The compositions and yield of flavonoid compounds of Polygonatum cyrtonema Hua (PC) are important indices of the quality of medicinal materials. However, the flavonoids compositions and accumulation mechanism are still unclear in PC. Here, we identified 22 flavonoids using widely-targeted metabolome analysis in 15 genotypes of PC. Then weighted gene co-expression network analysis based on 45 transcriptome samples was performed to construct 12 co-expressed modules, in which blue module highly correlated with flavonoids was identified. Furthermore, 4 feature genes including PcCHS1, PcCHI, PcCHS2 and PcCHR5 were identified from 94 hub genes in blue module via machine learning methods support vector machine-recursive feature elimination (SVM-RFE) and random forest (RF), and their functions on metabolic flux of flavonoids pathway were confirmed by tobacco transient expression system. Our findings identified representative flavonoids and key enzymes in PC that provided new insight for elite breeding rich in flavonoids, and thus will be beneficial for rapid development of great potential economic and medicinal value of PC.

Keywords: Biosynthesis; Flavonoid compounds; Machine learning; Metabolome; Polygonatum cyrtonema Hua.