Tsisal: Complete Deconvolution of DNA methylation data based on TOAST...

In TOAST: Tools for the analysis of heterogeneous tissues

Description

A function to conduct complete reference-free deconvolution on DNA methylation data. If a full reference or a partial reference panel is provided, this function also automatically annotate the solved proportions to known cell types.

Usage

Tsisal(Y_raw, K = NULL, knowRef = NULL, possibleCellNumber = 3:15)

Arguments

Y_raw	The DNA methylation 450K array data from complex tissues, rows for CpG sites and columns for samples.
K	The number of pure cell types, we allow users to pre-specify or use our method to estimate.
knowRef	The external reference panel for cell type label assignment.
possibleCellNumber	Range of possible number of cell types. Default is 3:15.

Value

estProp	Estimated proportions.
selMarker	Selected cell type-specific markers.
K	Optional number of cell types.

Author(s)

Weiwei Zhang <[e-mail]>

References

Complete deconvolution of DNA methylation signals from complex tissues: a geometric approach. Weiwei Zhang, Hao Wu and Ziyi Li.

Examples

### generate a simulation data
knowRef <- matrix(runif(5000*5), 5000, 5)
colnames(knowRef) <- paste0("CellType", 1:5)
Y_raw <- matrix(runif(5000*20), 5000, 20)
rownames(Y_raw) <- paste0("CpG", 1:5000)
colnames(Y_raw) <- paste0("Sample", 1:20)

Tsisal(Y_raw = Y_raw, K = 5, knowRef = knowRef)

## if cell type number is unknown
# Tsisal(Y.raw = Y_raw, K = NULL, knowRef = knowRef, possibleCellNumber = 4:10)

TOAST documentation built on Nov. 8, 2020, 5:55 p.m.