Run viewmastR using the new Rust implementation
viewmastR.RdThis function runs viewmastR using the new Rust implementation to generate various forms of training and test data from the given query and reference cell data sets.
Usage
viewmastR(
query_cds,
ref_cds,
ref_celldata_col,
query_celldata_col = NULL,
FUNC = c("mlr", "nn", "nb"),
norm_method = c("log", "binary", "size_only", "none"),
selected_genes = NULL,
train_frac = 0.8,
tf_idf = F,
scale = F,
hidden_layers = c(500, 100),
learning_rate = 0.001,
max_epochs = 10,
LSImethod = 1,
verbose = T,
backend = c("wgpu", "nd", "candle"),
threshold = NULL,
keras_model = NULL,
dir = "/tmp/sc_local",
return_probs = F,
return_type = c("object", "list"),
debug = F,
...
)Arguments
- query_cds
A cell data set (cds) to query.
- ref_cds
A reference cell data set (cds).
- ref_celldata_col
The column in the reference cell data set containing cell data.
- query_celldata_col
The column in the query cell data set containing cell data. Defaults to "viewmastR_pred" if not provided.
- FUNC
The machine learning model to use (multinomial linear regression - 'mlr' or multilayer neural network - 'nn'). Default is "mlr".
- norm_method
The normalization method to use. Options are "log", "binary", "size_only", or "none".
- selected_genes
A vector of pre-selected genes for analysis. If NULL, common features between reference and query are used.
- train_frac
The fraction of data to use for training. Default is 0.8.
- tf_idf
Boolean indicating whether to perform TF-IDF transformation. Default is FALSE.
- scale
Boolean indicating whether to scale the data. Default is FALSE.
- hidden_layers
A vector specifying the number of neurons in hidden layers for neural network models. Default is c(500, 100). To use one hidden layer supply a vector of length 1.
- learning_rate
The learning rate for training the model. Default is 1e-3.
- max_epochs
The maximum number of epochs for training the model. Default is 10.
- LSImethod
The method for Latent Semantic Indexing. Default is 1.
- verbose
Boolean indicating whether to display verbose output. Default is TRUE.
- dir
The directory to save output files. Default is "/tmp/sc_local".
- return_type
The type of output to return. Options are "object" or "list". Default is "object".
- debug
Run in debug mode
- ...
Additional arguments.