GENE ACTION OF MAIZE TRAITS AND RESISTANT SOURCE DETECTION AGAINST STALK ROT USING DEEP LEARNING TECHNIQUE

The early selection of pure lines against stalk rot and inheritance pattern of physiological and nutritional traits under disease stress is the main problem faced by the breeders in developing countries.One hundred maize lines were assessed for stalk rot using both artificial and natural inoculation.Six generations (P1, P2, F1, F2, BC1, and BC2) produced from the pure lines showing consistent resistance (Y11, EL7) and susceptible (DR59, DR69) responses during two growing seasons.Genetic images of these lines were obtained using DNA-based makers (SSRs) by performing polymerase chain reaction in the department of Medicinal Chemistry, University of Minnesota, USA.Studied traits showed the dominating impact of additive gene action in both seasons.Nutritional traits (protein, oil and starch contents) mostly exhibited negative correlation with lesion length while other showed non-significant correlation during spring and autumn.A deep learning model (Inception-V3) was trained with genetic images to distinguish between resistant and susceptible lines.The model showed 95% accuracy in resistant line detections.Considering the impact of the illness on physiological and nutritional traits, the current study favored phenotypic based selection in physiological traits.The study also showed the adverse impact of disease on nutritional traits.The current research not only aids in the management of the disease in high-yielding varieties, but also aids in identifying the resistant lines earlier against stalk rot using image processing model.