Get Signature Exposure from 'Signature' Object — get_sig

The expected number of mutations (or copy number segment records) with each signature was determined after a scaling transformation V ~ WH = W'H' where W' = WU' and H' = UH. The scaling matrix U is a KxK diagnal matrix (K is signature number, U' is the inverse of U) with the element corresponding to the L1-norm of column vectors of W (ie. the sum of the elements of the vector). As a result, the k-th row vector of the final matrix H' represents the absolute exposure (activity) of the k-th process across samples (e.g., for SBS, the estimated (or expected) number of mutations generated by the k-th process). Of note, for copy number signatures, only components of feature CN was used for calculating H'.

get_sig_exposure(
  Signature,
  type = c("absolute", "relative"),
  rel_threshold = 0.01
)

Arguments

Signature	a `Signature` object obtained either from sig_extract or sig_auto_extract, or just a raw exposure matrix with column representing samples (patients) and row representing signatures.
type	'absolute' for signature exposure and 'relative' for signature relative exposure.
rel_threshold	only used when type is 'relative', relative exposure less than (`<=`) this value will be set to 0 and thus all signature exposures may not sum to 1. This is similar to this argument in sig_fit.

Value

a data.table

References

Kim, Jaegil, et al. "Somatic ERCC2 mutations are associated with a distinct genomic signature in urothelial tumors." Nature genetics 48.6 (2016): 600.

Author

Shixiang Wang w_shixiang@163.com

Examples

# Load mutational signature
load(system.file("extdata", "toy_mutational_signature.RData",
  package = "sigminer", mustWork = TRUE
))
# Get signature exposure
expo1 <- get_sig_exposure(sig2)
expo1
expo2 <- get_sig_exposure(sig2, type = "relative")
expo2