AllelicImbalance (version 1.10.2)

detectAI: detectAI

Description

detection of AllelicImbalance

Usage

detectAI(x, ...)
"detectAI"(x, return.class = "DetectedAI", strand = "*", threshold.frequency = 0, threshold.count.sample = 1, threshold.delta.frequency = 0, threshold.pvalue = 0.05, inferGenotype = FALSE, random.ref = FALSE, function.test = "binom.test", verbose = TRUE, gc = FALSE, biasMatrix = FALSE)

Arguments

x

ASEset

...

internal arguments

return.class

class to return (atm only class 'logical')

strand

strand to infer from

threshold.frequency

least fraction to classify (see details)

threshold.count.sample

least amount of counts to try to infer allele

threshold.delta.frequency

minimum of frequency difference from 0.5 (or mapbias adjusted value)

threshold.pvalue

pvalue over this number will be filtered out

inferGenotype

infer genotypes based on count data in ASEset object

random.ref

set the reference as random if you dont know. Affects interpretation of results.

function.test

At the moment the only available option is 'binomial.test'

verbose

makes function more talkative

gc

use garbage collection when possible to save space

biasMatrix

use biasMatrix in ASEset, or use default expected frequency of 0.5 for all sites

Details

threshold.frequency is the least fraction needed to classify as bi tri or quad allelic SNPs. If 'all' then all of bi tri and quad allelic SNPs will use the same threshold. Everything under the treshold will be regarded as noise. 'all' will return a matrix with snps as rows and uni bi tri and quad will be columns. For this function Anything that will return TRUE for tri-allelicwill also return TRUE for uni and bi-allelic for the same SNP an Sample.

return.type 'ref' return only AI when reference allele is more expressed. 'alt' return only AI when alternative allele is more expressed or 'all' for both 'ref' and 'alt' alleles. Reference allele is the one present in the reference genome on the forward strand.

threshold.delta.frequency and function.test will use the value in mapBias(x) as expected value.

function.test will use the two most expressed alleles for testing. Make therefore sure there are no tri-allelic SNPs or somatic mutations among the SNPs in the ASEset.

inferGenotype(), set TRUE it should be used with as much samples as possible. If you split up the samples and run detectAI() on each sample separately, please make sure you have inferred the genotypes in before hand, alternatively used the genotypes detected by another variantCaller or chip-genotypes. Use ONLY biallelic genotypes.

Examples

#load example data
data(ASEset)
a <- ASEset

dai <- detectAI(a)