Table 3.
Results from PLS regression analysis
| SNP-selection strategy | Phenotype | # SNPs | C-V # PLS components* | R2 with 1 PLS component† | Average R2 per SNP‡ |
|---|---|---|---|---|---|
| SNPs from genes with ASKAT p value ≤ 0.05 | DBP-1 | 9414 | 1 | 0.418 | 7.1 × 10−5 |
| DBP-C | 10,337 | 0 | 0.424 | 6.2 × 10−5 | |
| SNPs from genes in Cytoscape pathway with FDR ≤ 0.05 | DBP-1 | 1239 | 0 | 0.333 | 4.4 × 10−4 |
| DBP-C | 866 | 1 | 0.302 | 5.2 × 10−4 | |
| SNPs from genes in Cytoscape common network | DBP-1 | 242 | 0 | 0.228 | 1.6 × 10−3 |
| DBP-C | 239 | 0 | 0.197 | 1.3 × 10−3 | |
*The optimal number of PLS components, as determined by 10-fold cross-validation.
†R2 values for a PLS model with 1 component.
‡The 1-component R2 divided by the number of SNPs in the PLS component with nonzero regression coefficient, which are 5920, 6835, 759, 580, 143, and 150 SNPs, respectively.