glm.enet: Variable selction based on elasticnet penalty.
In charliex210/sctools: Tools for single cell RNAseq analysis

Fits a multinomial model via penalized maximum likelihood.

1
2
3

glm.enet(mat, cell_labels, alpha = 0.5, lam = "lambda.min",
  family = "multinomial", type.multinomial = "grouped", nfolds = 10,
  cv.plot = FALSE, ...)

`mat`	Gene expression matrix, columns are cells and rows are genes.
`cell_labels`	cell clustering labels.
`alpha`	The elastic-net mixing parameter.
`lam`	Determine the lambda that according to cross-validation error - default is lambda.min. lambda.1se is also available.
`family`	Response type. see `glmnet`.
`type.multinomial`	If "grouped" then a grouped lasso penalty is used on the multinomial coefficients for a variable.
`nfolds`	number of folds - default is 10.
`cv.plot`	Save cross-validation plot.
`...`	Additional arguments passed on to `glmnet`.