Predict cell fractions and signature matrices with TAPE

Applies a trained TAPE model to processed bulk data, optionally with adaptive refinement in either overall or high-resolution mode.

Usage

tape_predict(
  model,
  test_x,
  genename,
  celltypes,
  samplename,
  adaptive = TRUE,
  mode = "overall",
  chunk_size = 16L
)

Arguments

model

A trained model returned by [tape_train()].

test_x

Numeric matrix with samples in rows and genes in columns.

genename

Character vector of gene names.

celltypes

Character vector of cell-type names.

samplename

Character vector of sample names.

adaptive

Logical scalar. Whether to run adaptive refinement.

mode

Character scalar. Either `"overall"` or `"high-resolution"`.

chunk_size

Integer scalar. Number of samples adapted simultaneously in high-resolution mode.

Value

A named list with components:

sigm

A named list of per-cell-type signature matrices in high-resolution mode, a signature matrix data frame in overall adaptive mode, and `NULL` when `adaptive` is `FALSE`.

pred

A data frame of predicted proportions with samples in rows and cell types in columns.

Details

High-resolution mode adapts every sample with its own copy of the trained model, and processes `chunk_size` samples concurrently as an ensemble of independent models. The supplied model is read only in this mode and is returned unmodified.

Examples

if (FALSE) { # \dontrun{
res <- tape_predict(
  model = model,
  test_x = processed$test_x,
  genename = processed$genename,
  celltypes = processed$celltypes,
  samplename = processed$samplename,
  mode = "high-resolution",
  chunk_size = 16L
)
} # }