tileHMM (version 1.0-7)

remove.short: Post-Processing of "tileHMM" Results

Description

Remove short regions that are likely to be spurious.

Usage

remove.short(regions, post, probe.pos, min.length = 1000,  min.score = 0.8, summary.fun = mean)

Arguments

regions

A matrix with information about the location of enriched regions.

post

A numeric vector with the posterior probability of ChIP enrichment for each probe.

probe.pos

A data frame with columns ‘chromosome’ and ‘position’ providing genomic coordinates for each probe.

min.length

Minimum length of enriched regions (see Details).

min.score

Minimum score for enriched regions (see Details).

summary.fun

Function used to summarise posterior probe probabilities into region scores.

Value

A matrix with two rows and one column for each remaining region.

Details

All regions that are shorter than min.length and have a score of less than min.score will be removed. To filter regions based on only one of these values set the other one to 0. Region scores are calculated based on posterior probe probabilities. The summary function used should accept a single numeric argument and return a numeric vector of length 1. If the probabilities in post are log transformed they will be transformed back to linear space before they are summarised for each region.

See Also

region.position

Examples

## create two state HMM with t distributions
state.names <- c("one","two")
transition <- c(0.1, 0.1)
location <- c(1, 2)
scale <- c(1, 1)
df <- c(4, 6)
model <- getHMM(list(a=transition, mu=location, sigma=scale, nu=df), 
    state.names)

## obtain observation sequence from model
obs <- sampleSeq(model, 500)

## make up some genomic probe coordinates
pos <- data.frame(chromosome = rep("chr1", times = 500), 
    position = seq(1, 18000, length = 500))

## calculate posterior probability for state "one"
post <- posterior(obs, model, log=FALSE)

## get sequence of individually most likely states
state.seq <- apply(post, 2, which.max)
state.seq <- states(model)[state.seq]

## find regions attributed to state "one"
reg.pos <- region.position(state.seq, region="one")

## remove short and unlikely regions
reg.pos2 <- remove.short(reg.pos, post, pos, min.length = 200, 
    min.score = 0.8)