train_VariSel: Title
In Marie-PerrotDockes/VariSel: Perform variable selection for different model types

Title

train_VariSel(
  X = NULL,
  Y,
  type,
  sepx = NULL,
  regressors = NULL,
  group = NULL,
  lambda = NULL,
  lambda1 = NULL,
  lambda2 = NULL,
  Sigma_12inv = NULL,
  type_S12_inv = "emp",
  p = NULL,
  q = NULL
)

`X`	a design matrix of explicatives variables for the linear model
`Y`	a matrix of responses
`type`	the type of models to fit. See details for a description of the different types.
`sepx`	if the type use fused lasso or group lasso on the columns of X it is the character separating the group value to the identifier values. For instance if the names of the columns of X are gene.allele and if we want to group variable that belong to the same gene use sepx=".".
`regressors`	if X is null and we are interested in the association between regressors and reponses in different group this is a matrix of regressors.
`group`	this is a charcater vector containing the different groups. It side must be the number of row of the matrix regressors. It indiciate for each rows in which group it is.
`lambda`	default to null. A numeric vector containing values for the regularisation parameter for lasso and group-lasso penalties.
`lambda1`	default to null. A numeric vector containing values for the lambda1 parameters (sparsity parmater for the fused lasso penalties).
`lambda2`	default to null. A numeric vector containing values for the lambda2 parameters (fusion parmater for the fused lasso penalties).
`Sigma_12inv`	a matrix of the square root of the inverse of the covariance matrix of the residuals. ( Use to remoove the dependance that may exist among the responses)
`type_S12_inv`	if Sigma_12inv is null it can be estimated using different type. See details for the differents types available.