runHMM: Copy number variation inference through a Hidden Markov Model

In CNOGpro: Copy Numbers of Genes in prokaryotes

Description

Implements a Viterbi algorithm for assigning most likely copy number to each chromosomal position in the chromosome.

Usage

runHMM(experiment, nstates = 5, changeprob = 1e-04, includeZeroState = T,
	errorRate = 0.001)

Arguments

experiment	An object of class CNOGpro
nstates	The possible number of states, not including state 0. The returned copy numbers will be in the range 0, 1, 2, ... , nstates
changeprob	The probability of transitioning from one state to another, used to set up the transition matrix.
includeZeroState	Whether or not to allow the copy number state 0 in the results
errorRate	The presumed fraction of erroneously mapped reads. Only needed when includeZeroState is set to TRUE. This numbers is used for setting the probability distribution of each observation in copy number state 0.

Details

For each read count observation the algorithm computes the probability of that observation in each possible state. The minimum path through the trellis is then calculated at the end.

Value

An object of class CNOGpro, with a HMMtable listing the breakpoints of different copy number states. The most probable states of each genetic element are also listed in the genes table of the object.

Author(s)

Ola Brynildsrud

Examples

data(carsonella)
carsonella_normalized <- normalizeGC(carsonella)
carsonella_hmm <- runHMM(carsonella_normalized)
plotCNOGpro(carsonella_hmm)

Example output

Loading required package: seqinr
Normalizing read counts for GC bias...
Overall values for chromosome  CP003467 :
Median read count:  66 
Median GC percentage:  0.17 
Calculating per-read weight based on GC percentage...
Successfully normalized chromosome  CP003467 
Running HMM... 
Analyzing chromosome CP003467 
Finished setting up transition and emission matrices. Starting Viterbi algorithm...
Attempting to create Viterbi matrix
Positions 1 - 18800 : Most probable state =  1 
Positions 18801 - 22000 : Most probable state =  2 
Positions 22001 - 94200 : Most probable state =  1 
Positions 94201 - 95200 : Most probable state =  0 
Positions 95201 - 174014 : Most probable state =  1 
Finished running Viterbi algorithm. Assigning most probable states to individual segments...
Copy number analysis complete. Use print to access results.

CNOGpro documentation built on May 2, 2019, 5:57 a.m.