robtest: Calls the robnb function to estimate the coefficients, and...

In DiPhiSeq: Robust Tests for Differential Dispersion and Differential Expression in RNA-Sequencing Data

Description

Calls the robnb function to estimate the coefficients, and then construct the statistical tests for DD and DE. It works for a single gene. y1 and y2 are count vectors for a single gene. diphiseq calls this function to do the calculation for each gene. Normal users often don't need to use this function directly.

Usage

robtest(y1, log.depth1, y2, log.depth2, c.tukey.beta = 4, c.tukey.phi = 4)

Arguments

y1	counts from group 1. a vector.
log.depth1	log(sequencing depths) for samples in group 1. a vector.
y2	counts from group 2. a vector.
log.depth2	log(sequencing depths) for samples in group 2. a vector.
c.tukey.beta	The c value for beta in Huber function. The default value, 4, is typically regarded as appropriate and should work for most datasets.
c.tukey.phi	The c value for phi in Huber function. The default value, 4, is typically regarded as appropriate and should work for most datasets.

Value

A vector that contains the elements: phi1: the estimated dispersion of sample group 1. phi2: the estimated dispersion of sample group 2. beta1: the estimated (log) expression of sample group 1. beta2: the estimated (log) expression of sample group 2. statistic.phi: the z statistic for DD. statistic.beta: the z statistic for DE. p.value.phi: the p value for DD. p.value.beta: the p value for DE.

Examples

d1 <- runif(10, min=1, max=2)
d2 <- runif(15, min=1, max=2)
y1 <- rnbinom(10, size=1, mu=d1*50)
y2 <- rnbinom(15, size=1, mu=d2*50)
res <- robtest(y1, log(d1), y2, log(d2))

DiPhiSeq documentation built on May 2, 2019, 11:56 a.m.