Abstract
Motivation Mapping quantitative trait loci (QTLs) is essential in plant and animal genetics, but identifying epistasis remains challenging. Machine learning methods, such as XGBoost with Shapley additive explanations (SHAP), can enhance QTL mapping by capturing complex relationships and evaluating feature importance.
Result SHAP-assisted XGBoost (SHAP-XGB) was compared to composite interval mapping (CIM), multiple interval mapping (MIM), inclusive CIM (ICIM), and BayesC using simulations and rice heading date data. SHAP-XGB performed comparably for main QTL effects and surpassed other methods in detecting epistatic interactions. SHAP’s ability to assess local importance improved the interpretation of marker interactions in plant and animal genetics.
Availability and Implementation R scripts for SHAP-XGB are available at https://github.com/Onogi/SHAP-XGB.
Contact onogiakio{at}gmail.com
Supplementary information Supplementary information, figures, and tables are available at the journal’s website.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
The main text was shortened for readability. Instead, Supplementary information was added.
