Molecular characterization of a distinct ginger chemotype from Thua Thien Hue, Vietnam, and the application of PCR-based markers for identifying unknown ginger populations in the region using machine learning

Abstract

Ginger (<i>Zingiber officinale</i> Roscoe) stands as an esteemed herbaceous spice due to its extensive applications in medical and culinary sectors. The variety of ginger indigenous to Thua Thien Hue, known as Hue's ginger, has long garnered recognition for its distinct aroma and unique oil composition. Regrettably, this ginger variety has intermingled with unidentified ginger types. Thus, the objective of this study is to identify DNA markers that can facilitate the identification of Hue's ginger. Such markers will enable the precise selection and preservation of the authentic ginger chemotype. To substantiate the distinctive genetic attributes of Hue's ginger, we employed two marker techniques: RAPD and <i>mat</i>K DNA barcoding. The RAPD technique demonstrated its robustness by generating an impressive number of 139 amplicons, with an absolute polymorphic rate of 100%. Among the resulting bands, two region-specific markers, OPA03-480 and OPB01-1150, were delineated for Hue's ginger. These specific markers facilitated the separation of Hue's ginger from other ginger chemotypes, shown by principal coordinates analysis. Furthermore, the alignment of the <i>mat</i>K gene sequence of Hue's ginger with the reference chloroplast genome substantiated the hypothesis that Hue's ginger possesses distinct genetic characteristics. This alignment revealed three transition variants within the matK gene of Hue’s ginger. Considering the extensive intermixing of ginger populations in Thua Thien Hue, we constructed an XGBoost machine-learning model using RAPD data to identify the most pivotal markers capable of effectively distinguishing between these populations. Our model identified OPN06-350, OPA03-480, OPD02-500, OPF04-950, and OPN06-300 as the most influential markers for population discrimination. This study not only furnishes molecular markers for the precise identification of a unique Vietnamese ginger chemotype but also advocates for the utilization of machine-learning methodologies employing PCR-based marker data for the identification of pivotal markers, a practice with promising implications for the effective differentiation of plant varieties in future endeavors.