test_my_model: Test your model

In scGate: Marker-Based Cell Type Purification for Single-Cell Sequencing Data

Test your model

Description

Wrapper for fast model testing on 3 sampled datasets

Usage

test_my_model(
  model,
  testing.version = "hsa.latest",
  custom.dataset = NULL,
  target = NULL,
  plot = TRUE
)

Arguments

model	scGate model in data.frame format
testing.version	Character indicating the version of testing tatasets to be used. By default "hsa-latest" will be used. It will be ignored if a custom dataset is provided (in Seurat format).
custom.dataset	Seurat object to be used as a testing dataset. For testing purposes, metadata seurat object must contain a column named 'cell_type' to be used as a gold standard. Also a set of positive targets must be provided in the target variable.
target	Positive target cell types. If default testing version is used this variable must be a character indicating one of the available target models ('immune','Lymphoid','Myeloid','Tcell','Bcell','CD8T','CD4T', 'NK','MoMacDC','Plasma_cell','PanBcell'). If a custom dataset is provided in Seurat format, this variable must be a vector of positive cell types in your data. The last case also require that such labels were named as in your cell_type meta.data column.
plot	Whether to return plots to device

Value

Returns performance metrics for the benchmarking datasets, and optionally plots of the predicted cell type labels in reduced dimensionality space.

Examples

scGate.model.db <- get_scGateDB()
# Browse the list of models and select one:
model.panBcell <-  scGate.model.db$human$generic$PanBcell
# Test the model with available testing datasets
panBcell.performance <- test_my_model(model.panBcell, target = "PanBcell")
model.Myeloid <-  scGate.model.db$human$generic$Myeloid
myeloid.performance <- test_my_model(model.Myeloid, target = "Myeloid")
     

scGate documentation built on May 29, 2024, 2:56 a.m.