RunLCV: LCV

In jean997/causeSims: Generate Summary Statistic Data and Analyze with a Variety of MR Methods

Description

Main function for running LCV. Obtained from https://github.com/lukejoconnor/LCV

Usage

RunLCV(
  ell,
  z.1,
  z.2,
  no.blocks = 100,
  crosstrait.intercept = 1,
  ldsc.intercept = 1,
  weights = 1/pmax(1, ell),
  sig.threshold = .Machine$integer.max,
  n.1 = 1,
  n.2 = 1,
  intercept.12 = 0
)

Arguments

ell	Mx1 vector of LD scores;
z.1	Mx1 vector of estimated marginal per-normalized-genotype effects on trait 1 (or Z scores; invariant to scaling);
z.2	Mx2 vector of effects on trait2;
no.blocks,	number of jackknife blocks;
crosstrait.intercept	0 if crosstrait LDSC intercept should be fixed and 1 otherwise;
ldsc.intercept	0 if LDSC intercept should be fixed and 1 otherwise (1 is recommended);
weights	Mx1 vector of regression weights;
sig.threshold,	threshold above which to discard chisq statistics for the purpose of estimating the LDSC intercept; large-effect SNPs discarded above sig_threshold*mean(z.x^2);
intercept.12	covariance between sampling errors for Z1 and Z2, only needed if crosstrait_intercept=0 If the 2 GWAS are disjoint, this can be set to zero.
n1	sample size for trait 1, only needed if ldsc.intercept=1;
n2	sample size for trait 2, only needed if ldsc.intercept=1;

jean997/causeSims documentation built on Sept. 4, 2020, 4:29 p.m.