BPfam: Family function for beta-Poisson generalized linear model

In nghiavtr/BPSC: Beta-Poisson Model for Single-Cell RNA-seq Data Analyses

Description

Family function for beta-Poisson generalized linear model

Usage

BPfam(alp, bet, lam1, lam2, link = "log")

Arguments

alp	A non-negative value, the parameter of Beta function (alpha)
bet	A non-negative value, the parameter of Beta function (beta)
lam1	A non-negative value, the parameter for scaling
lam2	A non-negative value, the parameter for smoothing, used for BP4 or BP5
link	The link function, as a character string, similar to quasi-poisson family, typically log link

Value

An object of class "family" for function glm() to fit a beta-Poisson generalized linear model

Examples

set.seed(2015)
control.dat=rBP(100,alp=0.6,bet=1.5,lam1=20,lam2=0.05)
treated.dat=rBP(100,alp=0.2,bet=5,lam1=200,lam2=0.1)
x=c(control.dat,treated.dat)
y=c(rep(1,length(control.dat)),rep(2,length(treated.dat)))
fdat=data.frame(x=x,group=y)
fam0=do.call("BPfam", list(alp=0.6,bet=1.5,lam1=20,lam2=0.05, link = "log"))
fit=glm(x~.,data=fdat,family=fam0)
fit
summary(fit)$coefficients

nghiavtr/BPSC documentation built on May 23, 2019, 4:42 p.m.