SEARCH

SEARCH BY CITATION

Disclaimer: Supplementary materials have been peer-reviewed but not copyedited.

FilenameFormatSizeDescription
gepi21777-sup-0001-SuppMat.zip508K

Figure S1. Gene score trend results, plotting cross-validated MSE from within the validation set against the number of top single nucleotide polymorphisms (SNPs) in the score for (i) un-weighted (UW) scores, (ii) univariate (UV) weighted scores and (iii) multivariate (MV) weighted scores for (a) high-density lipoprotein (HDL) in Whitehall II Study (WH-II), (b) low-density lipoprotein (LDL) in British Women's Health and Heart Study (BWHHS), (c) HDL in BWHHS.

Figure S2. Mean square error (MSE) prediction results for Hyper-Lasso, varying the penalty parameter along the x-axis and ranging over different values of the shape parameter in the normal exponential gamma (NEG) prior: (i) 10 (ii) 5, (iii) 2.5, (iv) 1, (v) 0.75, (vi) 0.5, (vii) 0.1 and (viii) 0.05, for (a) high-density lipoprotein (HDL) in Whitehall II Study (WH-II), (b) low-density lipoprotein (LDL) in British Women's Health and Heart Study (BWHHS) and (c) HDL in BWHHS, each with all single nucleotide polymorphisms (SNPs) included in the model.

Figure S3. Overall minimum mean square error (MSE) for results within the validation set from Ridge Regression, Lasso and Hyper-Lasso with shape parameters {10, 5, 2.5, 1, 0.75, 0.5, 0.1, 0.05} in the normal exponential gamma (NEG) prior, ranging over different pre-filtered subsets of (i) all, the top (ii) 1,000, (iii) 100, (iv) 75, (v) 50, (vi) 25, (vii) 15, (viii) 10 and (ix) 5 single nucleotide polymorphisms (SNPs) input into the model (plotted on log scale) for (a) high-density lipoprotein (HDL) in Whitehall II Study (WH-II), (b) low-density lipoprotein (LDL) in British Women's Health and Heart Study (BWHHS), (c) HDL in BWHHS.

Figure S4. Relationship between mean square error (MSE) and Pearson correlation within the validation set for cross-validated gene scores predicting low-density lipoprotein (LDL) in Whitehall II Study (WH-II) using (a) univariate weights and (b) multivariate weights.

Figure S5. Multivariate weighted gene score trend results, plotting mean square error (MSE) against the number of top single nucleotide polymorphisms (SNPs) in the score for predicting high-density lipoprotein (HDL) either (i) within Whitehall II Study (WH-II) using cross-validation, (ii) within British Women's Health and Heart Study (BWHHS) using cross-validation, (iii) from WH-II as the training set onto BWHHS as the test set and (iv) from BWHHS as the training set onto WH-II as the test set.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.