GPOSYSH: Genomic Prediction of Oryza sativa Yield and Subpopulation Using Hybrid Methods.

Abstract

Accurate prediction of breeding values is challenging due to the genotype-phenotype relationship is crucial and necessary for producing crops with elite genotypes. This paper is about investigating and predicting the phenotypic trait Height and Yeild in a genotype.

Background: Most of the existing studies focus on genetic methods or Machine learning models, in this, we implemented a hybrid combination of genetic methods and machine learning models that accurately predicted phenotypic trait yield, height and subpopulation.

Methodology: Our proposed methodology for genomic prediction of yield in Oryza sativa (rice) involves a two-level classification approach. First, we classify biological sequences and cluster them using the UPGMA algorithm on a phylogenetic tree. Then, we use advanced machine learning techniques like Random Forest, and K-Nearest Neighbours to predict GEBVs with 85-95% accuracy on rice subpopulations.

Results: we achieved an accuracy of 93% when compared with other stated literature in this paper.

Conclusion: This approach overcomes limitations and effectively enhances crop breeding by capturing the genotype-phenotype relationship.