call_genotypes: Call genotypes based on EM algorithm

In Yi-Jiang/MethyGenotyper: MethylGenotyper: Accurate estimation of SNP genotypes from DNA methylation data

Call genotypes based on EM algorithm

Description

The Expectation–maximization (EM) algorithm is used to fit a mixture of three beta distributions representing the three genotypes (AA, AB, and BB) and one uniform distribution representing the outliers (adapted from ewastools). Probe-specific weights were used in the EM algorithm.

Usage

call_genotypes(
  RAI,
  pop,
  type,
  maxiter = 50,
  bayesian = FALSE,
  platform = "EPIC",
  verbose = 1
)

Arguments

RAI	A matrix of RAI (Ratio of Alternative allele Intensity) for probes. Provide probes as rows and samples as columns.
pop	Population to be used to extract AFs. One of EAS, AMR, AFR, EUR, SAS, and ALL.
type	One of snp_probe, typeI_probe, and typeII_probe.
maxiter	Maximal number of iterations for the EM algorithm.
bayesian	Use the Bayesian approach to calculate posterior genotype probabilities.
platform	EPIC or 450K.
verbose	Verbose mode: 0/1/2.

Value

A list containing

RAI	Ratio of Alternative allele Intensity
shapes	Shapes of the mixed beta distributions
weights	Prior probabilities that the RAI values belong to one of the three genotypes
U	Overall probability of RAI values being outlier
outliers	Probability of each RAI value being outlier
logLik	Log-likelihood
GP	Genotype probabilities of the three genotypes

Yi-Jiang/MethyGenotyper documentation built on Sept. 4, 2024, 12:47 p.m.