Rcppreqtl Manual
User Manual:
Open the PDF directly: View PDF 
.
Page Count: 9
Package ‘rcppreqtl’
September 12, 2018
Type Package
Title eqtl optimization using 'RcppEigen'
Version 0.99.0
Date 2016-01-19 The 'rcppreqtl' package uses 'RcppEigen' and C++11
numeric solver for eqtl optimization.
Author Vasyl Zhabotynsky [aut, cre], Wei Sun [aut]
Maintainer Vasyl Zhabotynsky <vasyl@unc.edu>
Description
Analysis of combined total and allele specific reads from the human trio experiment using RNA-
seq data.
License GPL (>= 2) | file LICENSE
Depends R (>= 2.15.1),MASS,VGAM
LazyLoad yes
LinkingTo RcppEigen, Rcpp
Imports Matrix (>= 1.1-0), RcppEigen (>= 0.3.2.0), Rcpp (>= 0.11.0),
stats, utils
NeedsCompilation yes
Archs i386, x64
Rtopics documented:
fit .............................................. 2
fitsh ............................................. 3
makeXmatr ......................................... 4
readCounts ......................................... 5
simu2 ............................................ 6
simu4 ............................................ 7
Index 9
1
2fit
fit Optimization wrapper, maximizing the joint model of total (TReC) and
allele specific (ASE) counts for autosomes
Description
Performs optimization of joint TReC and ASE for autosome and tests with lrt test for two hypothe-
ses hypotheses: additive and parent of origin.
Usage
fit(subset=NULL, data, traceit=FALSE)
Arguments
subset a subset of entries to be tested, but default is set to NULL which leads to fitting
all the genes in the table
data an object of class readCounts including read counts and necessary supporting
information
traceit include more debug output, by default set to FALSE
Value
a list of following matrices:
full matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the full model fit as well as appended two p-value tests for additive and parent
of origin effect
testadd matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the restricted to b0=0 model fit
testpoo matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the restricted to b1=0 model fit
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fitsh,data,simu4,simu2,readCounts.
Examples
## Not run:
# fitting autosome data for a full model with allele-specific counts collected on gene level:
percase = 0.1
dblcnt = 0.2
mn = 100
b0 = 0
b1 = 0
phiNB = .5;
phiBB=phiNB/4
fitsh 3
niter = 10
betas = c(3,.2,.05,.5)
ss=2
dep = makeXmatr(ss)
dat = simu4(num=niter, Xmatr=dep$Xmatr, haplotype=dep$thp, totmean=mn, percase=percase, dblcnt=dblcnt, phiNB=phiNB, phiBB=phiBB, b0=b0, b1=b1, betas=betas)
fit(subset=NULL, data=dat, traceit=FALSE)
## End(Not run)
fitsh Optimization wrapper, maximizing the joint model of total (TReC) and
allele specific (ASE) counts for autosomes
Description
Performs optimization of joint TReC and ASE for autosome and tests with lrt test for two hypothe-
ses hypotheses: additive and parent of origin. This function modification assumes that phiNB and
phiBB are common for the gene of interest. Otherwise the model is exactly the same.
Usage
fitsh(subset=NULL, data, traceit=FALSE)
Arguments
subset a subset of entries to be tested, but default is set to NULL which leads to fitting
all the genes in the table
data an object of class readCounts including read counts and necessary supporting
information
traceit include more debug output, by default set to FALSE
Value
a list of following matrices:
full matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the full model fit as well as appended two p-value tests for additive and parent
of origin effect
testadd matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the restricted to b0=0 model fit
testpoo matrix with columns: log(phiNB) and -log(phiBB), coefficients, -log(likelihood)
for the restricted to b1=0 model fit
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fitsh,data,simu2,simu4,readCounts.
4makeXmatr
Examples
## Not run:
# fitting autosome data for a full model with allele-specific counts collected on gene level:
percase = 0.1
dblcnt = 0.2
mn = 100
b0 = 0
b1 = 0
phiNB = .5;
phiBB=phiNB/4
niter = 10
betas = c(3,.2,.05,.5)
ss=2
dep = makeXmatr(ss)
dat = simu4(num=niter, Xmatr=dep$Xmatr, haplotype=dep$thp, totmean=mn, percase=percase, dblcnt=dblcnt, phiNB=phiNB, phiBB=phiBB, b0=b0, b1=b1, betas=betas)
fitsh(subset=NULL, data=dat, traceit=FALSE)
## End(Not run)
makeXmatr Create example design matrix for simulations
Description
Produces a design matrix of several sample sizes which can be used to generate simulated dataset
Usage
makeXmatr(ss)
Arguments
ss Sample size class: ss=1 implies dample size 32, ss=2 implies sample size 64,
etc
Value
a design matrix of 4 variables
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fitsh,data,simu4,simu2,readCounts.
readCounts 5
Examples
## Not run:
# fitting autosome data for a full model with allele-specific counts collected on gene level:
percase = 0.1
dblcnt = 0.2
mn = 100
b0 = 0
b1 = 0
phiNB = .5;
phiBB=phiNB/4
niter = 100
betas = c(3,.2,.05,.5)
ss=2
dep = makeXmatr(ss)
## End(Not run)
readCounts A list object that should be used as input to optimization fit or fitsh
function.
Description
It should contain at least total read counts (TReC), overall allele-specific counts (ASE), paternal
allele-specific counts and haplotype classification 0 to 4. Also should include X matrix with covari-
ates such as intercept, library depth, principal components.
Value
haplotype a matrix defining the haplotype status of the individual for each gene - AB=0,
BA=1, AA=2, BB=3 coded as PaternalMaternal haplotypes. Each row - one
gene
trc matrix of TReC counts. Each row - one gene
asn matrix of ASE counts for subject (column) for corresponding genes (row).
asnp matrix of ASE counts belonging to paternal allele, for correponding subjects and
genes as in asn.
haplotypeA haplotype modification for a setup with allele-specific reads collected for multi-
ple SNPs in a gene. This block by default is NULL, but can be used by simula-
tion function to study scenarios when reads are collected not on gene level, but
on SNP level
asnA allele-specific count corresponding to haplotypeA
asnpA paternal allele-specific count corresponding to haplotypeA
Xdesign matrix for total read counts - a place to include intercept, library depth,
other covariates such as batch effect
params if data is produced by simulation function parameter values can be stored here
for further comparisons
settings other important settings for the simulation such as mean of total read counts,
percentage of allele specific counts, percentage of reads double-counted with
neighboring SNPs can be stored here
6simu2
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fit,fitsh,simu2,simu4.
Examples
## Not run:
# see total read counts (TReC) for first 2 X chromosome genes of a data example:
rc = readCounts(haplotype=haplotypef, trc=trc, asn=asnf, asnp=asnpf, haplotypeA=haplotyped, asnA=asnm, asnpA=asnpm,
X=Xmatr, params=c(phiNB, phiBB, b0, b1, betas), settings=c(totmean, percase, dblcnt))
## End(Not run)
simu2 Simulate a dataset in a format acceptable by a fit function
Description
Creates an object of a class readCounts with data simulated according to a provided setup
Usage
simu2(num, Xmatr, haplotype, totmean, percase=0.1, dblcnt=0, phiNB=1, phiBB=0.5, b0=0, b1=0, betas=rep(1,4))
fullest = fit(subset=NULL, data=dat, traceit=FALSE)
Arguments
num number of iterations
Xmatr design matrix for total read counts covariates
haplotype classes of haplotypes 0 - AA, 1 - AB, 2 - BA, 3 - BB where first letter represents
paternal allele
totmean average total gene expression
percase percentage of reads classified as allele-specific, default value 10%
dblcnt optional output considering a simulation split into 2 SNPs with double-counting.
Default value 0.
phiNB over-dispersion for Negative-Binomial distribution, default value 1
phiBB over-dispersion for Beta-Binomial distribution, default value 0.5
b0 additive eQTL, default value 0
b1 parent of origin effect, default value 0
betas covariates for design matrix Xmatr
Value
an object of class readCounts
simulated dataset that can be used to fit the model
simu4 7
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fitsh,data,simu4,simu2,readCounts.
Examples
## Not run:
# fitting autosome data for a full model with allele-specific counts collected on gene level:
percase = 0.1
dblcnt = 0.2
mn = 100
b0 = 0
b1 = 0
phiNB = .5;
phiBB=phiNB/4
niter = 100
betas = c(3,.2,.05,.5)
ss=2
dep = makeXmatr(ss)
dat = simu2(num=niter, Xmatr=dep$Xmatr, haplotype=dep$thp, totmean=mn, percase=percase, dblcnt=dblcnt, phiNB=phiNB, phiBB=phiBB, b0=b0, b1=b1, betas=betas)
## End(Not run)
simu4 Simulate a dataset in a format acceptable by a fit function
Description
Creates an object of a class readCounts with data simulated according to a provided setup
Usage
simu4(num, Xmatr, haplotype, totmean, percase=0.1, dblcnt=0, phiNB=1, phiBB=0.5, b0=0, b1=0, betas=rep(1,4))
fullest = fit(subset=NULL, data=dat, traceit=FALSE)
Arguments
num number of iterations
Xmatr design matrix for total read counts covariates
haplotype classes of haplotypes 0 - AA, 1 - AB, 2 - BA, 3 - BB where first letter represents
paternal allele
totmean average total gene expression
percase percentage of reads classified as allele-specific, default value 10%
dblcnt optional output considering a simulation split into 4 SNPs with double-counting.
Default value 0.
phiNB over-dispersion for Negative-Binomial distribution, default value 1
8simu4
phiBB over-dispersion for Beta-Binomial distribution, default value 0.5
b0 additive eQTL, default value 0
b1 parent of origin effect, default value 0
betas covariates for design matrix Xmatr
Value
an object of class readCounts
simulated dataset that can be used to fit the model
Author(s)
Vasyl Zhabotynsky vasyl@unc.edu
See Also
fitsh,data,simu4,simu2,readCounts.
Examples
## Not run:
# fitting autosome data for a full model with allele-specific counts collected on gene level:
percase = 0.1
dblcnt = 0.2
mn = 100
b0 = 0
b1 = 0
phiNB = .5;
phiBB=phiNB/4
niter = 100
betas = c(3,.2,.05,.5)
ss=2
dep = makeXmatr(ss)
dat = simu4(num=niter, Xmatr=dep$Xmatr, haplotype=dep$thp, totmean=mn, percase=percase, dblcnt=dblcnt, phiNB=phiNB, phiBB=phiBB, b0=b0, b1=b1, betas=betas)
## End(Not run)
