Learn R Programming

cancerTiming (version 3.1.8)

eventTiming: Estimate the time of events in tumor data

Description

Estimate the proportion of time spent between different chromosomal abnormalities based on the allele frequencies of mutated locations.

Usage

eventTiming (x, m, history, totalCopy, method = c("fullMLE","partialMLE", "Bayes"), type = c("gain", "CNLOH"), seqError = 0, bootstrapCI = NULL, B = if (method =="Bayes") 10000 else 500, CILevel = 0.95, normCont = 0, verbose = TRUE, returnAssignments = FALSE, coverageCutoff = 1, minMutations = 10, init = NULL, maxiter = 100, tol= 1e-04, mutationId = 1:length(x),...)

Arguments

x
vector. the number of reads/fragments containing the variant

m
vector. the number of reads/fragments covering the location with the variant (the coverage)

history
a matrix, based on the history of the region (see Details)

totalCopy
integer. the total number of copies of the tumor DNA for this region

method
what estimation method to use, one of ``fullMLE",``partialMLE",``Bayes"
type
type of region, either a gain or a CNLOH region

seqError
Probability of sequencing error
bootstrapCI
type of bootstrap confidence interval to calculate, one of ``parametric", ``nonparametric". If NULL, then the confidence interval is not calculated

B
number of bootstrap samples to take (or simulations from the posterior for Bayesian estimation)

CILevel
At what level the confidence intervals should be calculated.
normCont
the proportion of normal contamination, between 0 and 1.

verbose
logical. Whether to give additional warnings as the program is running.

returnAssignments
logical. Whether to return the probabilistic assignments of mutations to allele frequencies generated by the EM algorithm. Also returns the x,m values for those that pass the filter.

coverageCutoff
minimum value for m[i]; any entries with m[i]

minMutations
minimum number of mutations required.
init
initial value of multinomial parameter q passed to estimateQ.

maxiter
maximum number of iterations in calculation q.

tol
tolerance in the convergence of q

mutationId
identification values of the mutations (vector of the same length as x and m). Default is indexing values, 1:length(x). Used when returnAssignments=TRUE so that the assignments of the mutations to allele frequencies can be linked with the original mutations if there has been filtering in eventTiming, e.g. due to depth of coverage.
...
Arguments passed to internal fitting function for Bayesian Method. For example, `alpha' gives the Dirichlet prior of the bayesian estimate (default=1), `tdf' gives the number of degrees of freedom for the t proposal density used in the bayesian estimate (default=4), `bayesApproxMethod' gives the method for calculating the approximate distribution (default is ``sir"; ``inv" is for K=1 when the problem is 1-dimensional and one can easily grid and get the approximate posterior density and cdf).

Value

A list with values
pi
estimate of pi vector
piCI
bootstrap confidence interval, if requested
q
estimate of the multinomial parameter q
perLocationProb
output from estimateQ giving per location P(P[i]|X[i],q), if requested. Only locations used in the estimation are included.
optimDetails
optimization details from estimateQ
call
list of the parameters of the call to the function: history, totalCopy, type, exactAllele, normCont, coverageCutoff, minMutations. In addition, `alleleSet' is included in this list, which is the set of possible alleles for this history, after adjusting for normal contamination.

References

Durinck S, et al. (2011). ``Temporal Dissection of Tumorigenesis in Primary Cancers." Cancer Discovery, 1(2): 137--143. Greenman CD, et al. (2012). ``Estimation of rearrangement phylogeny for cancer genomes." Genome Research, 22(2): 346--361. Purdom E, et al. (2013). ``Timing Chromosomal Abnormalities within Cancer Samples.'' Bioinformatics, 29(24): 3113--3120.

Examples

Run this code
data(mutData)
ACNLOH<-matrix(c(1,3,1,0),ncol=2,nrow=2,byrow=TRUE)
onlyMuts<-subset(mutData,is.na(rsID) & position <= 1.8E7)
onlyMuts$t_depth<-onlyMuts$t_ref_count+onlyMuts$t_alt_count
x<-eventTiming(x=onlyMuts$t_alt_count,m=onlyMuts$t_depth,
    history=ACNLOH,totalCopy=2,type="CNLOH",normCont=0.22)
x$pi #estimate of time of stages
x$q #estimate of the multinomial (likelihood of each of the alleles)
x$call$alleleSet #possible set of alleles after 
                #adjusting for normal contamination

Run the code above in your browser using DataLab