Extended Data Table 4: Polygenic risk profiling and binomial sign tests.
| pT | Polygenic risk profiling | Binomial sign test | ||
|---|---|---|---|---|
|
| ||||
| r 2 | P | No. SNPs (%) | P | |
| 0.000001 | 0.000715 | 0.0174 | 3 (100) | 0.125 |
| 0.00001 | 8.40E-05 | 0.415 | 12 (66.7) | 0.194 |
| 0.0001 | 2.57E-05 | 0.652 | 62 (58.1) | 0.126 |
| 0.001 | 5.87E-06 | 0.829 | 481 (53.6) | 0.0605 |
| 0.01 | 8.67E-05 | 0.407 | 3632 (51.1) | 0.101 |
| 0.1 | 0.00142 | 0.000797 | 26106 (50.4) | 0.126 |
| 0.2 | 0.00126 | 0.00156 | 45166 (50.6) | 0.00331 |
| 0.3 | 0.00116 | 0.00246 | 61074 (50.5) | 0.00627 |
| 0.4 | 0.00125 | 0.00168 | 74676 (50.5) | 0.00335 |
| 0.5 | 0.0011 | 0.00317 | 86429 (50.4) | 0.00758 |
| 1 | 0.000924 | 0.00684 | 124361 (50.3) | 0.0116 |
The table shows the predictive value of a Psychiatric Genetics Consortium (PGC) MDD-trained polygenic risk score on the CONVERGE results. Predictive values are shown at varying p-value thresholds (pT) from P<=1×10−6 to 1 (i.e. all results). P is the P-value of the prediction and r2 the amount of variance explained (thus the table shows that including all SNPs from PGC-MDD explained 0.09% of MDD risk in CONVERGE.). The number of independent SNPs at each threshold is presented (No. SNPs); the significance of the observed fraction (%) demonstrating a consistent direction of effect was assessed by a one-sided binomial sign test.