A new method for exploring gene–gene and gene–environment interactions in GWAS with tree ensemble methods and SHAP values

BMC Bioinformatics

Megan E. Cahill¹  Andrew Thomas DeWan²  Mette Langaas³  Signe Riemer-Sørensen⁴  Pål V. Johnsen⁵ 

[1] Department of Chronic Disease Epidemiology and Center for Perinatal, Pediatric and Environmental Epidemiology, Yale School of Public Health, 1 Church Street, 06510, New Haven, CT, USA;Department of Chronic Disease Epidemiology and Center for Perinatal, Pediatric and Environmental Epidemiology, Yale School of Public Health, 1 Church Street, 06510, New Haven, CT, USA;Gemini Center for Sepsis Research, Department of Circulation and Medical Imaging, NTNU, Norwegian University of Science and Technology, Prinsesse Kristinas gate 3, 7030, Trondheim, Norway;Department of Mathematical Sciences, Norwegian University of Science and Technology, A. Getz vei 1, 7491, Trondheim, Norway;SINTEF DIGITAL, Forskningsveien 1, 0373, Oslo, Norway;SINTEF DIGITAL, Forskningsveien 1, 0373, Oslo, Norway;Department of Mathematical Sciences, Norwegian University of Science and Technology, A. Getz vei 1, 7491, Trondheim, Norway;

关键词: GWAS;    Tree ensemble models;    XGBoost;    SHAP;    Model explainability;    Gene–gene and gene–environment interactions;   

DOI  :  10.1186/s12859-021-04041-7

来源: Springer

【 摘 要 】

BackgroundThe identification of gene–gene and gene–environment interactions in genome-wide association studies is challenging due to the unknown nature of the interactions and the overwhelmingly large number of possible combinations. Parametric regression models are suitable to look for prespecified interactions. Nonparametric models such as tree ensemble models, with the ability to detect any unspecified interaction, have previously been difficult to interpret. However, with the development of methods for model explainability, it is now possible to interpret tree ensemble models efficiently and with a strong theoretical basis.ResultsWe propose a tree ensemble- and SHAP-based method for identifying as well as interpreting potential gene–gene and gene–environment interactions on large-scale biobank data. A set of independent cross-validation runs are used to implicitly investigate the whole genome. We apply and evaluate the method using data from the UK Biobank with obesity as the phenotype. The results are in line with previous research on obesity as we identify top SNPs previously associated with obesity. We further demonstrate how to interpret and visualize interaction candidates.ConclusionsThe new method identifies interaction candidates otherwise not detected with parametric regression models. However, further research is needed to evaluate the uncertainties of these candidates. The method can be applied to large-scale biobanks with high-dimensional data.

【 授权许可】

CC BY