Implements the MRLocus slope fitting step, in which the estimated
coefficients and their original SE are used to determine
the mediation slope (alpha
), and the dispersion of individual
signal clusters around the slope (sigma
). This function
follows the colocalization step fitBetaColoc
and extractForSlope
.
The output fitSlope
can be visualized with
plotMrlocus
. For details on the model,
see Supplementary Methods of the MRLocus manuscript.
See vignette for example of model interpretation.
fitSlope(
res,
sd_beta = NULL,
mu_alpha = NULL,
sd_alpha = NULL,
sd_sigma = NULL,
...
)
list with the following named elements:
beta_hat_a
- point estimates of coefficients for A from colocalization
beta_hat_b
- " " for B
sd_a
- sampling SD for beta_hat_a
(in practice original
SE are provided here)
sd_b
- " " for beta_hat_b
" "
alleles (optional) data.frame with allele information
prior SD for beta A (default value will be derived from data)
prior mean for alpha
(default value of 0)
prior SD for alpha
(default value will be derived from data)
prior SD for sigma
(default value of 1)
further arguments passed to rstan::sampling
list with the following elements: stanfit
object,
original estimated coefficients and standard deviations,
as well as the alleles
data.frame (if it was provided)
Note that since v0.0.26, the default prior mean for alpha is 0, whereas in the paper it was derived from the data.
Note that if summary statistics for only one SNP are provided a warning will be printed (this is not a recommended use of MRLocus) and a parametric simulation is used to estimate the slope, instead of the Bayesian model.
Anqi Zhu*, Nana Matoba*, Emma P. Wilson, Amanda L. Tapia, Yun Li, Joseph G. Ibrahim, Jason L. Stein, Michael I. Love. MRLocus: identifying causal genes mediating a trait through Bayesian estimation of allelic heterogeneity. (2021) PLOS Genetics https://doi.org/10.1371/journal.pgen.1009455