LD score regression

Source: R/ldsc.R

LD score regression

snp_ldsc(
  ld_score,
  ld_size,
  chi2,
  sample_size,
  blocks = 200,
  intercept = NULL,
  chi2_thr1 = 30,
  chi2_thr2 = Inf,
  ncores = 1
)

snp_ldsc2(
  corr,
  df_beta,
  blocks = NULL,
  intercept = 1,
  ncores = 1,
  ind.beta = cols_along(corr),
  chi2_thr1 = 30,
  chi2_thr2 = Inf
)

Arguments

ld_score

Vector of LD scores.

ld_size

Number of variants used to compute ld_score.

chi2

Vector of chi-squared statistics.

sample_size

Sample size of GWAS corresponding to chi-squared statistics. Possibly a vector, or just a single value.

blocks

Either a single number specifying the number of blocks, or a vector of integers specifying the block number of each chi2 value. Default is 200 for snp_ldsc(), dividing into 200 blocks of approximately equal size. NULL can also be used to skip estimating standard errors, which is the default for snp_ldsc2().

intercept

You can constrain the intercept to some value (e.g. 1). Default is NULL in snp_ldsc() (the intercept is estimated) and is 1 in snp_ldsc2() (the intercept is fixed to 1). This is equivalent to parameter --intercept-h2.

chi2_thr1

Threshold on chi2 in step 1. Default is 30. This is equivalent to parameter --two-step.

chi2_thr2

Threshold on chi2 in step 2. Default is Inf (none).

ncores

Number of cores used. Default doesn't use parallelism. You may use bigstatsr::nb_cores().

corr

Sparse correlation matrix. Can also be an SFBM.

df_beta

A data frame with 3 columns:

  • $beta: effect size estimates

  • $beta_se: standard errors of effect size estimates

  • $n_eff: either GWAS sample size(s) when estimating beta for a continuous trait, or in the case of a binary trait, this is 4 / (1 / n_control + 1 / n_case); in the case of a meta-analysis, you should sum the effective sample sizes of each study instead of using the total numbers of cases and controls, see doi:10.1016/j.biopsych.2022.05.029 ; when using a mixed model, the effective sample size needs to be adjusted as well, see doi:10.1016/j.xhgg.2022.100136 .

ind.beta

Indices in corr corresponding to df_beta. Default is all.

Value

Vector of 4 values (or only the first 2 if blocks = NULL):

  • [["int"]]: LDSC regression intercept,

  • [["int_se"]]: SE of this intercept,

  • [["h2"]]: LDSC regression estimate of (SNP) heritability (also see coef_to_liab),

  • [["h2_se"]]: SE of this heritability estimate.

Examples

bigsnp <- snp_attachExtdata()
G <- bigsnp$genotypes
y <- bigsnp$fam$affection - 1
corr <- snp_cor(G, ind.col = 1:1000)

gwas <- big_univLogReg(G, y, ind.col = 1:1000)
df_beta <- data.frame(beta = gwas$estim, beta_se = gwas$std.err,
                      n_eff = 4 / (1 / sum(y == 0) + 1 / sum(y == 1)))

snp_ldsc2(corr, df_beta)
#>       int        h2 
#> 1.0000000 0.2335429 
snp_ldsc2(corr, df_beta, blocks = 20, intercept = NULL)
#>       int    int_se        h2     h2_se 
#> 0.4986445 0.2526338 0.6195226 0.1818980