Modeling of Flowering Time in Vigna radiata with Artificial Image Objects, Convolutional Neural Network and Random...

Area of Science:

• Agricultural Science
• Genetics
• Machine Learning

Background:

• Flowering time is crucial for crop adaptation and breeding new varieties.
• Developing accurate predictive models for flowering time is essential for agricultural advancements.
• Mung bean breeding requires strategies to adapt to changing environmental conditions.

Purpose of the Study:

• To develop a novel approach for predicting mung bean flowering time.
• To identify key genetic and environmental factors influencing flowering time.
• To enhance breeding programs for developing varieties with desired flowering traits.

Main Methods:

• Utilized single nucleotide polymorphism (SNP) markers and weather data.
• Encoded genotypic and weather data into artificial image objects.
• Constructed a convolutional neural network (CNN) model for prediction.
• Employed saliency maps and Score-CAM for feature importance analysis.

Main Results:

• The CNN model accurately predicted mung bean flowering time.
• The model effectively used limited weather data (5 days before to 20 days after planting).
• Identified critical SNP markers and weather patterns influencing flowering time.

Conclusions:

• The proposed CNN approach offers a highly accurate method for flowering time prediction in mung beans.
• This method aids breeders in leveraging genotypic and phenotypic diversity for targeted variety development.
• The findings support the creation of mung bean varieties better suited to specific environmental conditions.