snapCGH (version 1.42.0)

runBioHMM: This function implements the BioHMM

Description

This function reads in a dataset of log2 ratios and the corresponding clone and covariate information. It calculates a heterogeneous HMM when there are 1,2,3,4 or 5 underlying states and chooses between them using either the AIC or BIC. It then assigns clones using a modified version of the Viterbi algorithm.

Usage

runBioHMM(input, useCloneDists = TRUE, covariates, criteria="AIC", delta=NA
,var.fixed=FALSE, epsilon = 1e-06, numiter = 30000)

Arguments

input

An object of class MAListor SegList

useCloneDists

Boolean stating whether the distance between clones should be incorportated into the HMM. If false then the HMM becomes homogeneous.

covariates

This is a dataframe containing information about covariate factors. The first two columns should be Chrom (giving the chromosome on which a clone is located) and Mb (giving the position of the chromosome along a particular chromosome in Megabases). The order should be the same as that described above with the following crucial difference. No covariate information about the first clone is used in the segmentation. Hence, for each chromosome, there should be one less row in the covariate dataframe than in the datainfo dataframe corresponding to this missing chromosome. This is important if the transition matrix is to be calculated correctly.

criteria

Options are AIC or BIC depending upon which we want to use to distinguish between the number of states

delta

A variable to be assigned if the BIC is used.

var.fixed

Logical variable - TRUE if you want to tie the variance to be the same across all states. Defaults to FALSE

epsilon

Stopping criterion for the optimization algorithm.

numiter

Number of iterations to be used in the optimization algorithm.

Value

The model returns an object of class SegList.

References

Marioni, J.C., Thorne, N.P., Tavar\'e, S., BioHMM: a heterogeneous Hidden Markov Model for segmenting array CGH data, submitted