run_bayesNMF: execute bayesNMF for signature deconvolution

In jburos/mutsigNMF:

Description

execute bayesNMF for signature deconvolution

Usage

run_bayesNMF(mat, tumor.type = "tumor.type", max_k = 10, n_runs = 10,
  iter_per_run = 1e+05, tol = 1e-07, a0 = 10, phi = 1, prior = "L1KL",
  hyper = FALSE)

Arguments

mat	(matrix) N*M matrix of ints with N mutation frequencies for each of M samples
max_k	(int) max hypothesized number of signatures to look for. default: 10
n_runs	(int) number of times to run optimization code. default: 10
iter_per_run	(int) number of iter for each run. default: 100000
tol	(float) tolerance used to assess convergence
a0	(int) value of hyperparameter 'a'. default: 10
phi	(int) value of hyperparameter 'phi'. default: 1 (appropriate for l1-loss with exponential prior)
prior	(chr) either L1KL (expoential priors) or L2KL (half-normal priors). default: L1KL
hyper	(logical) if TRUE, reduce the effect of hyper-mutant samples in the signature discovery. default: FALSE
tumor_name	(chr) name of tumor, used for plotting headers, etc. default: 'tumor.type'

Value

list of signatures, where each signature is a named list containing plots & data

jburos/mutsigNMF documentation built on May 18, 2019, 9:19 p.m.