GWASLab provides a simple function to calculate the variance explained by each SNP.
!!! tip "Available since v3.4.20"
mysumstats.get_per_snp_r2()
It needs effect size BETA and effect allele frequency EAF for calculation.
If N is available, it will also calculate the F-statistics.
.get_per_snp_r2() options |
DataType | Description | Default |
|---|---|---|---|
mode |
q or b |
q: quantitative trait; b: binary trait |
q |
For quantitative traits (mode="q"), GWASLab will use BETA, EAF to calculate SNPR2.
.get_per_snp_r2() options |
DataType | Description | Default |
|---|---|---|---|
vary |
float or se |
Var(Y); if vary="se",Var(Y) will be estimated using SE |
1 |
k |
int |
k for calculating F | 1 |
!!! info "For SNP i"
When `vary=1` and k=1:
$$ h^2_{i} = 2 \times \beta_i^2 \times EAF \times (1 - EAF) $$
$$ F = h^2_{i} \times (n -2) / (1 - h^2_{i}) $$
For binary traits (mode="b"), ncase, ncontrol and prevalence are needed to estimate the variance of liability explained by variants:
.get_per_snp_r2() options |
DataType | Description | Default |
|---|---|---|---|
ncase |
int |
number of cases | - |
ncontrol |
int |
number of controls | - |
prevalence |
float |
prevalence in general population | - |
!!! quote
Equation 10 in Lee, S. H., Goddard, M. E., Wray, N. R., & Visscher, P. M. (2012). A better coefficient of determination for genetic profile analysis. Genetic epidemiology, 36(3), 214-224.
- Implementation adopted from TwoSampleMR https://rdrr.io/github/MRCIEU/TwoSampleMR/src/R/add_rsq.r
!!! example ``` SNPID EAF BETA N STATUS 0 1:725932_G_A 0.9960 -0.0737 166718 9999999 1 1:725933_A_G 0.0040 0.0737 166718 9999999 2 1:737801_T_C 0.0051 0.0490 166718 9999999 3 1:749963_T_TAA 0.8374 0.0213 166718 9999999 4 1:751343_T_A 0.8593 0.0172 166718 9999999
mysumstats.get_per_snp_r2()
Mon Jul 17 01:39:02 2023 Start to calculate per-SNP heritability...
Mon Jul 17 01:39:02 2023 -Calculating per-SNP rsq by 2 * (BETA**2) * AF * (1-AF) / Var(y)...
Mon Jul 17 01:39:02 2023 -Var(y) is provided: 1...
Mon Jul 17 01:39:02 2023 -Calculating F-statistic: F = [(N-k-1)/k] * (r2/1-r2)... where k = 1
Mon Jul 17 01:39:02 2023 -For r2, SNPR2 is used.
Mon Jul 17 01:39:02 2023 Finished calculating per-SNP heritibility!
SNPID EAF BETA N STATUS SNPR2 F
0 1:725932_G_A 0.9960 -0.0737 166718 9999999 0.000043 7.215732
1 1:725933_A_G 0.0040 0.0737 166718 9999999 0.000043 7.215732
2 1:737801_T_C 0.0051 0.0490 166718 9999999 0.000024 4.062184
3 1:749963_T_TAA 0.8374 0.0213 166718 9999999 0.000124 20.600305
4 1:751343_T_A 0.8593 0.0172 166718 9999999 0.000072 11.927080
```