Graduate Students Seminar

Speaker: Upama Paul Chowdhury

Title

Robust Genomic Prediction and Heritability Estimation using Density Power Divergence

Abstract

Genomic prediction involves using DNA information, such as genetic markers, to estimate the traits or performance of plants and animals, enabling breeders to make more informed selection decisions. This work focuses on developing a robust statistical framework for genomic prediction using single nucleotide polymorphism (SNP) data from plant and animal breeding, including multi-field trials. The proposed approach integrates all estimated genetic marker effects into the modeling process, addressing the challenge of high-dimensional data while preserving essential biological information. A novel aspect of this framework is the use of one-stage and two-stage linear mixed models combined with the robust minimum density power divergence estimator (MDPDE) to accurately estimate genetic effects on phenotypic traits. Through simulation studies and applications to both artificial datasets and a real-world maize breeding dataset, the proposed method demonstrates superior predictive accuracy and heritability estimation compared to existing techniques. Notably, the MDPDE-based approach shows strong resilience to data contamination, underscoring its potential for enhancing the reliability and robustness of genomic prediction in practical breeding programs.