Machine learning-based identification of elite genotypes in the endangered Nilgirianthus ciliatus through qualitative and quantitative trait analysis

Abstract

Nilgirianthus ciliatus is an economically valuable endangered medicinal plant with a significant influence on traditional medicine and Ayurveda formulation. Its rarity in natural habitats precludes scientific investigation into its potential medicinal and other industrial applications. The current study examined the qualitative, quantitative and machine learning (ML) predictions for the identification of elite genotypes of N. ciliatus in India’s Western Ghats. The gas chromatography-mass spectroscopy (GC–MS) revealed the presence of betazole, neophytadiene, hexadecanoic acid methyl ester, octadecanoic acid, and squalene. The genotype NC 10 was found to yield high squalene content (793.0 ng), while the highest α-glucosidase Inhibitory Activity was shown by NC 2. The artificial neural network (ANN) demonstrated a high prediction accuracy (MSE value = 2.43E-02 while R value = 0.99992) in both the training and the testing sets of data. Genetic markers produced 140 bands, out of which 115 were polymorphic (82.14 %). Further, NC 10, NC 8, and NC 6 elite genotypes of N. ciliatus from three distinct agroclimatic zones were commended as industrially significant high-yielding characteristics and determined to be best suitable for cultivation. This study would serve as a foundation for understanding the use of artificial neural networks in elite genotype selection for efficient secondary metabolite synthesis.