RunJGNsc framework main function

RunJGNsc framework main function

RunJGNsc(
  observed.list,
  warm = 1000,
  iter = 5000,
  mask.rate = 0.15,
  nrep = 50,
  min.cell = 3,
  runNetwork = F,
  l1.vec = NULL,
  l2.vec = NULL,
  a1 = 3,
  b1 = 1,
  dropThreshold = 0.75
)

Arguments

observed.list	the list containing the matrices of K conditions. dim: genes by samples
warm	the number of warm steps in the MCMC procedure
iter	the number of iterations in the MCMC procedure
mask.rate	iterative imputation procedure
nrep	number of iterations in the imputation procedure, default is 50
min.cell	the min number of cells that should a gene should express
runNetwork	logical. Run the joint graphical lasso procedure or not.
l1.vec	if runNetwork=T, the vector of candidate values for the tuning parameter lambda1
l2.vec	if runNetwork=T, the vector of candidate values for the tuning parameter lambda2
a1	alpha parameter for Beta(alpha,beta) distribution which is the prior for non-dropouts. default is 3. The expected non-dropout rate is a1/(a1+b1) = 0.75 by default.
b1	beta parameter for Beta(alpha,beta) distribution which is the prior for non-dropouts. default is 1.

Value

theta.star.npn: the imputed and Gaussian transformed list of matrices

if runNetwork = T, then it will return the JGL results (the precision matrices can be accessed by result$JGL$theta) , aic.table (the AIC values with corresponding tuning parameter candidates), partcorr (the list of partial correlation matrices)