Learn R Programming

ebGenotyping (version 2.0.1)

mstep: CM steps

Description

This function calculates the CM steps of ECM algorithm for the model described in 'An Empirical Bayes Method for Genotyping and SNP detection Using Multi-sample Next-generation Sequencing Data'.

Usage

mstep(mu0, delta0, zm1, z0, zp1, dat, cvg, eps = 1e-06)

Arguments

mu0
a vetor of the same length as number of positions: the initial value of position effect mu.
delta0
a vetor of the same length as number of samples: the initial value of sample effect delta.
zm1
the output of estep: the posterior probabilities of genotype RR for n samples at m positions
z0
the output of estep: the posterior probabilities of genotype RV for n samples at m positions
zp1
the output of estep: the posterior probabilities of genotype VV for n samples at m positions
dat
a n*m matrix: the ith row, jth column of the matrix represents the non-reference counts of ith sample at jth position.
cvg
a n*m matrix: the ith row, jth column of the matrix represents the depth of ith sample at jth position.
eps
a single value: a threshold to control the convergence criterion. The default is 1e-06.

Value

mu
the optimal value of mu in current CM steps
delta
the optimal value of delta in current CM steps
pRR
the optimal value of the probability of RR in current CM steps
pRV
the optimal value of the probability of RV in current CM steps

Details

The value of mu and delta must satisfy that each element of outer(delta,mu,"+") must less than zero. This is the requirement of the model described in 'An Empirical Bayes Method for Genotyping and SNP detection Using Multi-sample Next-generation Sequencing Data'.

References

Na You and Gongyi Huang.(2016) An Empirical Bayes Method for Genotyping and SNP detection Using Multi-sample Next-generation Sequencing Data.