Fig 4.tif (1017.09 kB)

Comparison of polygenic risk prediction methods for 13 complex diseases.

figure

posted on 2016-12-30, 18:26 authored by Jianxin Shi, Ju-Hyun Park, Jubao Duan, Sonja T. Berndt, Winton Moy, Kai Yu, Lei Song, William Wheeler, Xing Hua, Debra Silverman, Montserrat Garcia-Closas, Chao Agnes Hsiung, Jonine D. Figueroa, Victoria K. Cortessis, Núria Malats, Margaret R. Karagas, Paolo Vineis, I-Shou Chang, Dongxin Lin, Baosen Zhou, Adeline Seow, Keitaro Matsuo, Yun-Chul Hong, Neil E. Caporaso, Brian Wolpin, Eric Jacobs, Gloria M. Petersen, Alison P. Klein, Donghui Li, Harvey Risch, Alan R. Sanders, Li Hsu, Robert E. Schoen, Hermann Brenner, Rachael Stolzenberg-Solomon, Pablo Gejman, Qing Lan, Nathaniel Rothman, Laufey T. Amundadottir, Maria Teresa Landi, Douglas F. Levinson, Stephen J. Chanock, Nilanjan Chatterjee

For all figures, the y-coordinate is the prediction R² in the observational scale. “1D” denotes 1D PRS; “2D, blood eSNPs” denotes 2D PRS using blood eSNPs as high-prior SNP set. In the x-axis, “NO” denotes PRS without winner’s curse correction; “LASSO” and “MLE” denote lasso-type and MLE winner’s curse correction, respectively. (A) Prediction R² values for six diseases in WTCCC data, estimated based on five-fold cross-validation. (B) Prediction R² values for three GWAS of cancers, estimated based on ten-fold cross-validation. (C) Prediction R² values for four complex diseases estimated based on independent validation samples.