mgcca: Generalized canonical correlation with missing individuals

In isglobal-brge/GCCA: Generalized Canonical Correlation Analysis having missing individuals

Description

Generalized canonical correlation with missing individuals

Usage

mgcca(
  x,
  nfac = 2,
  scale = TRUE,
  pval = TRUE,
  scores = FALSE,
  method = "solve",
  lambda,
  mc.cores = 1,
  ...
)

Arguments

x	list of matrices. Each matrix should have the ids in the rownames. Missing is not allowed (see details)
nfac	...
scale	...
pval	should p-values of correlation between variables and shared canonical variates be computed? Default is TRUE.
scores	should canonical variables be computed for each table? Default is FALSE. See details
method	...
lambda	...
mc.cores	...

Details

eigen with Rfast hd.eigen select number of components speed up process (RSpectra?) rownames matrices ... si hay distintos ordena - si no, no. missings ... dos opciones#' scores .. si queremos correlation with each table! inversa ... "solve" para n>>p (pero lento) (solve function) "penalized" penalized adding lambda*I (rfunctions cgls). Requires lambda "geninv" n<p generalized inverse (rfunctions geninv) "ginv" generalized inverse (MASS ginv)

Value

a list consisting of

Y	canonical components for the shared space
corsY	correlation between variables and shared canonical components
scores	canonical componets for each table
p.values	p-values of correlation between variables and shared canonical components
AVE	indicators of model quality based on the Average Variance Explained (AVE): AVE for each table, AVE_outer (average accross tables), AVE_inner(for the shared component).

Examples

see vignette

isglobal-brge/GCCA documentation built on Feb. 19, 2022, 9:21 a.m.