CFITTransfer: Transfer the learned gene expression signature onto a new...
In pengminshi/cFIT: cFIT (common Factor Space Integration & Transfer)

Estimate the data specific parameters through Iterative Nonnegative Matrix Factorization (iNMF), by minimizing the objective function

1/n||X -(HW^TΛ + 1_n b^T)||_F^2

CFITTransfer(
  Xtarget,
  Wref,
  max.niter = 100,
  tol = 1e-05,
  init = NULL,
  future.plan = c("multicore", "sequential", "transparent", "multisession", "cluster"),
  workers = parallel::detectCores() - 1,
  seed = 0,
  verbose = T
)

`Xtarget`	An ncells-by-ngenes gene expression matrix
`Wref`	An ngenes-by-r reference low dimensional factor matrix
`max.niter`	integer, max number of iterations (default 100).
`tol`	numeric scalar, tolerance used in stopping criteria (default 1e-5).
`init`	a list of parameters for parameter initialization: lambda, b, H
`future.plan`	plan for future parallel computation, can be chosen from 'sequential','transparent','multicore','multisession' and 'cluster'. Note that Rstudio does not support 'multicore'.
`workers`	additional parameter for `future::plan()`, in cases of 'multicore','multisession' and 'cluster'.
`seed`	random seed used (default 0)
`verbose`	boolean scalar, whether to show extensive program logs (default TRUE)

a list containing

H: estimated factor loading matrix of size ncells-by-r
b: estimated shift vector of size p (ngenes)
lambda: estimated scaling vector of size p (ngenes)
convergence: boolean, whether the algorithm converge
obj: numeric scalar, value of the objective function at convergence or when maximum iteration achieved
niter: integer, the iteration at convergence (or maximum iteration if not converge)
delta: numeric scalar, the relative difference in last objective update
params: list of parameters used for the algorithm: max.iter, tol, nrep