trainScSimilarity: Trains scRNA-seq data via tunable logistic regression model

In zktuong/kelvinny: Kelvinny The Incredible!

Trains scRNA-seq data via tunable logistic regression model

Description

Trains scRNA-seq data via tunable logistic regression model

Usage

trainScSimilarity(
  train_data,
  train_cell_type,
  test_data,
  train_genes = NULL,
  standardize = TRUE,
  nfolds = 10,
  a = 0.9,
  l.min = FALSE,
  multinomial = FALSE,
  nParallel = parallel::detectCores(),
  ...
)

Arguments

train_data	Seurat object, SummarizedExperiment object or expression matrix for training
train_cell_type	The cell types/clusters in the training data set
test_data	Seurat object, SummarizedExperiment/SingleCellExperiment object or expression matrix for testing later
train_genes	Genes to use for training. If not provided, it will try to pick from all genes in the training dataset as per default glmnet.
standardize	a logical value specifying whether or not to standardize the train matrix
nfolds	integer specifying bin for cross validation. Use all samples if doing LOOCV.
a	tunable regularization parameter. 0 = ridge (L2), 1 = LASSO (L1), in between = Elastic-net
l.min	logical. Choose between lambda.min or lambda.1se
multinomial	logical. Choose between family = 'binomial' or 'multinomial'.
nParallel	integer specifying number of cores for parallelization.
...	other functions pass to glmnet

Value

Generates a trained model for predicting cell types for scRNAseq data

Examples

fit <- trainScSimilarity(trainData, clusters, testData)

zktuong/kelvinny documentation built on July 18, 2024, 10:42 a.m.