predictTilState: Predict state of tumor-infiltrating CD8 T-cells from...

In carmonalab/TILPRED: Tumor-infiltrating CD8+ T-cell state predictor for scRNA-seq data

Predict state of tumor-infiltrating CD8 T-cells from single-cell signature enrichment scores

Description

predictTilState This function evaluates a logistic regression model to predict the state of individual CD8 tumor-infiltrating lymphocytes (mouse or human) based on their transcriptomes (scRNA-seq data)

Usage

predictTilState(
  data,
  nCores = 1,
  human = F,
  scoreThreshold = 0.5,
  cellCycleThreshold = 0.2,
  filterCD8T = T
)

Arguments

data	single-cell expression object of CD8 T-cells of class SingleCellExperiment. For gene expression matrix, only gene expression ranks in each cell will be used and therefore any cell-to-cell normalization method used equivalent (as long as gene ranks are conserved among the top ~5"%" genes). E.g. UMI counts, CPM, TPM, TMM are equivalent input types. Gene names correspond to mouse gene symbols (e.g. Cncb2).
nCores	number of cores used for (AUCell) AUC scores computation (Default 1)
human	logical value indicating if input matrix correspond to human genes (by default mouse data is expected)
scoreThreshold	probability threshold [0,1] for assigning cell states. If all state probabilities are below this threshold, 'unknown' state is assigned. Default 0.5
cellCycleThreshold	probability threshold [0,1] for assigning cell cyling state; Values in the range [0.1,0.2] are recommended (Default 0.2)
filterCD8T	automatic pre-filter CD8 T cells before classifing CD8 T cell states (Default TRUE). If filterCD8T is set to FALSE, all input cells are assumed to be CD8 T cells, even though they migh show no features of this cell type.

Value

a two-element list containing 1) predictedState, the predicted states for CD8 T cells (naive, effector memory, exhausted, memoryLike, or "unknown" if no class had a score above a threshold of scoreThreshold); or the predicted cell type for non CD8 T cells: Treg (Foxp3 Regulatory T cells), CD4T (non Treg CD4+ T cells), NKT (NK T cells), Tcell_unknown (T cells of other kinds) and Non-Tcell (for cell types other than T cells, e.g. Myeloid, B cells, NKs); 2) stateProbabilityMatrix, a matrix of number_of_cells x number_of_states (4) of probabilities of cell c belonging to class s, only for CD8 T cells; 3) cycling, logical vector indicating for each cell whethere there is a high cell cycle signal (independent to the cellular sub-type/state signal), and 4) cyclingScore AUC score for the cell cycle signature

Examples

data(B16CD8TIL_SCE)
x <- predictTilState(data=B16CD8TIL_SCE)
table(x$predictedState)
head(x$stateProbabilityMatrix)
head(x$cyclingScore)

carmonalab/TILPRED documentation built on Feb. 12, 2024, 8:34 p.m.