generate_feature

file

The initial value for p in Homozygous Beta-Binomial model, default is 0.999

hom_p

The initial value for p in Heterozygous Beta-Binomial model, default is 0.5

het_p

The initial value for rho in Homozygous Beta-Binomial model, default is 0.005

hom_rho

The initial value for rho in Heterozygous Beta-Binomial model, default is 0.1

het_rho

A vector of whether the sample is contaminated: 0 for pure; 1 for contaminated

mixture

Cutoff allele frequency value between hom and high, default is 0.99

homcut

Cutoff allele frequency value between high and het, default is 0.7

highcut

Cutoff allele frequency value between het and low, default is 0.3

hetcut

Generates features from each pair of input VCF objects for training contamination detection model.

Imports Variant Calling Format file into R. It can detect
whether a sample contains contaminant from the same species.
In the first stage of the approach, a change-point detection
method is used to identify copy number variations for filtering.
Next, features are extracted from the data for a support vector
machine model. For log-likelihood calculation, the deviation
parameter is estimated by maximum likelihood method. Using
a radial basis function kernel support vector machine, the
contamination of a sample can be detected.

generate_feature: Feature Generation for Contamination Detection Model

Description

Usage

Arguments

Value