This function takes as input individual phenotype and genotype data
and performs BMS for a feature-SNP pair (or a SNP). The output
includes posterior probabilities of different types of GxT
interactions as well as the log marginal likelihood, which can be
used for optimizing the hyperparameter values by an empirical Bayes
approach. BMS is performed by either MCMC followed by bridge
sampling or MAP estimation followed by Laplace approximation. In
the latter case, a relative marginal likelihood is returned. That
is, the log marginal likelihood value is shifted by a
constant. Note that this shift does not affect the posterior
probabilities or hyperparameter optimization.
tu.lambda must be supplied when modeling a random effect.
do_bms(
data,
fn,
rint = FALSE,
method,
p.m = NULL,
phi,
phi0 = sqrt(1000),
kappa = 0.002,
nu = 0.002,
kappa.u = 0.002,
nu.u = 0.002,
tu.lambda = NULL,
summary = TRUE,
seed = 1,
n.cores = 4
)A list containing vectors of subjects, genotypes, phenotypes, and treatment indicators, which must be named "subject", "g", "y", and "t", respectively.
A character string specifying the function. This must be one of "nonlinear" and "linear", corresponding to nonlinear and linear models, respectively.
A logical as to whether the phenotypes need to be
RINT-transformed. This cannot be set to TRUE if fn is
set to "nonlinear".
A character string spcifying the method for parameter estimation and computation of the marginal likelihood. This must be eigher "mcmc.bs", for MCMC followed by bridge sampling, or "map.lap", for MAP estimation followed by Laplace approximation.
A vector of hyperparameters of the model prior.
A vector of hyperparameters of the effect prior.
A scalar specifying the hyperparameter on the intercept.
A scalar specifying the hyperparameter of the gamma prior on the residual error precision.
A scalar specifying the hyperparameter of the gamma prior on the residual error precision.
A scalar specifying the hyperparameter of the gamma prior on the random intercept.
A scalar specifying the hyperparameter of the gamma prior on the random intercept.
A list obtained from
get_tu_lambda. It must contain the transposed
eigen vector matrix and the eigen values of the covariance
matrix. The element names must be "tU" and "lambda".
A Boolean variable indicating whether a summary or
MCMC samples should be stored. This is only applicable when
method is set to "mcmc.bs".
An integer specifying a seed for RNG.
An integer specifying the number of cores.
A list object containing:
fn - A character string specifying the function.
ranef - A logical.
rint - A logical as to whether the phenotypes have
been RINT-transformed.
p.m - A vector of hyperparameters of the model prior.
seed - A integer specifying a seed fo RNG.
ln.p.y.given.m - A named vector of the log marginal
likelihood given each model. If method is set to
"map.lap", the value is shifted by a constant.
ln.p.y - A scalar value of the log marginal
likelihood. If method is set to "map.lap", the value is
shifted by a constant.
p.m.given.y - A named vector of posterior probability of the
models.
ml.errors - A named vector of errors in the log marginal
likelihood. This element is included only when method is
set to "mcmc.bs".
stan.list - A list of stanfit objects or data frames
containing MCMC results for all eight models. This element is inclued only when
method is set to "mcmc.bs". If summary is set to
TRUE, a list of data frames containing posterior summary
is returned. If the option is set to FALSE, a list of
stanfit objects is returned.
optim.list - A list of outputs from the optim
function from the stat package containing MAP
estimates and Hessian. This element is included only when
method is set to "map.lap".