extract_signatures_initialiser: Use optimization to generate initial parameter values for...

In kgori/sigfit: Flexible Bayesian inference of mutational signatures

Description

Use optimization to generate initial parameter values for MCMC sampling

Usage

extract_signatures_initialiser(
  counts,
  nsignatures,
  model = "multinomial",
  opportunities = NULL,
  sig_prior = NULL,
  exp_prior = NULL,
  chains = 1,
  ...
)

Arguments

counts	Numeric matrix of observed mutation counts, with one row per sample and one column per mutation type.
nsignatures	Integer or integer vector indicating the number(s) of signatures to extract.
model	Name of the model to sample from. Admits character values "multinomial" (default), "poisson", "negbin", "normal", "nmf" (an alias for "multinomial"), and "emu" (an alias for "poisson").
opportunities	Numeric matrix of optional mutational opportunities, with one row per sample and one column per mutation type. It also admits character values "human-genome" or "human-exome", in which case the mutational opportunities of the reference human genome/exome will be used for every sample.
sig_prior	Numeric matrix with one row per signature and one column per mutation type, to be used as the Dirichlet priors for the mutational signatures. Only used when a single value is provided for nsignatures. Default priors are uniform.
chains	Integer indicating number of chains to be initialised (default is 1).
...	Additional arguments to be passed to rstan::optimizing.

Value

List of initial values to be passed to extract_signatures via the init argument.

kgori/sigfit documentation built on Feb. 3, 2022, 12:04 p.m.