fitHMM: Fit an HMM

In quevedor2/WadingPool: A toolkit for shallow/low-pass whole genome sequencing analysis

Description

Runs depmixs4 to fit an HMM to a set of observations given an expected number of states. It will then configure the emission dataframe to be compatible with downstream analysis

Usage

fitHMM(
  sample,
  het_cnt,
  states = 2,
  family = poisson(),
  nstart = 10,
  tiles = NULL,
  is.sim = FALSE,
  ret.raw = FALSE,
  multi = FALSE,
  tr_switch = c(0, 0.2),
  tr_same = c(0.9, 0.95)
)

Arguments

sample	String containing sample ID in columns
het_cnt	Dataframe for sample observations
states	Integer for expected number of states
family	Family of distribution to use (default=poisson())
nstart	Number of starts to try out for EM (Default=10)
tiles	GenomicRanges object for corresponding chrs and bins. If none is given, it will fill in the genomic poisition randomly (Default=NULL)
is.sim	Boolean flag for whether data is simulation or not (Default=FALSE)
ret.raw	Boolean flag to return the fitted model depmixS4 object
multi	Boolean whether to use depmixs4::multistart for fitting model
tr_switch	numeric vector length 2: min/max probabilty range of transition to different state
tr_same	numeric vector length 2: min/max probabilty range of transition to same state

Value

Dataframe containing the emission proabbility for each state and a fitted HMM

quevedor2/WadingPool documentation built on Dec. 22, 2021, 10:59 a.m.