# rmRNAseq v0.1.0

0

0th

Percentile

## RNA-Seq Analysis for Repeated-Measures Data

A differential expression analysis method for RNA-seq data from repeated-measures design using general linear model framework and parametric bootstrap inference. The method accounts for the dependence of gene expression levels due to the repeated-measures through continuous autoregressive correlation structure. The method is described in Chapter 4 of Nguyen (2018) <https://lib.dr.iastate.edu/cgi/viewcontent.cgi?article=7433&context=etd>.

# rmRNAseq

The goal of rmRNAseq is to conduct differential expression analysis using RNA-seq data from repeated-measures designs, where mRNA samples are obtained repeatedly at different times from same experimental units. Our method is developed based on a general linear model framework with continuous autoregressive correlation structure of order one, accompanied by a parametric bootstrap inference strategy to conduct general hypothesis testings.

## Installation

You can install the released version of rmRNAseq from CRAN with:

install.packages("rmRNAseq")


(The package has just been submitted to CRAN on June 26, 2019; it usually takes about 10 days to receive their feedback.)

And the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("ntyet/rmRNAseq")


## Example

This is a basic example which shows you how to solve a common problem:

library(rmRNAseq)
data(dat)
data(design)
data(covset)
Subject <- covset$ear # identity of experimental units Time <- covset$time # times at which mRNA samples are taken
Nboot <- 2  # for real data analysis, use Nboot at least 100
ncores <- 1 # for real data analysis and if the computer allows, increase ncores to save time
print.progress <- FALSE
saveboot <- FALSE
counts <- dat[1:3,]
C.matrix <- list()
# test for Line main effect
C.matrix[[1]] <- limma::makeContrasts(line2, levels = design)
# test for Time main effect
C.matrix[[2]] <- limma::makeContrasts(time2, time6, time24, levels = design)
names(C.matrix) <- c("line2", "time")
TCout <- rmRNAseq:::TC_CAR1(counts, design, Subject, Time, C.matrix,
Nboot, ncores, print.progress, saveboot)
#> running bootstrap sample nrep =  1
#> ---------------------------------------------
#> running bootstrap sample nrep =  2
#> ---------------------------------------------
names(TCout)
#> [1] "NewTime"    "TimeMinOut" "ori.res"    "pqvalue"
TCout$NewTime[1:4] #> [1] 0.1492027 0.8028973 1.0000000 0.0000000 TCout$pqvalue$pv #> line2 time #> 1 0.2857143 0.1428571 #> 2 0.1428571 0.1428571 #> 3 1.0000000 0.1428571 TCout$pqvalue\$qv
#>       line2      time
#> 1 0.4285714 0.1428571
#> 2 0.4285714 0.1428571
#> 3 1.0000000 0.1428571


## Functions in rmRNAseq

 Name Description estimate.m0 Estimate Number of True Null Hypotheses Using Histogram-based Method design Reparameterized Design Matrix NewTimeEst Estimate New Time Points covset Covariate Set Associated with RFI RNA-seq TimeMin Identify Time Points Mapping to 0 and 1 DESeq2Fit Analysis of LPS RFI RNA-seq data Using DESeq2 TC_CAR1_sc A Wrap Function to analyze a Simulated Data - All Cases glsCAR1 Fit General Linear Model with corCAR1 Correlation Structure for One Gene dat RFI RNA-seq Data pauc_out Evaluation of Differential Expression Analysis Methods sc_Symm Simulating Count Data From The Output of Real Data Analysis (corSymm) voomlimmaFit Analysis of RFI RNA-seq data Using voom glsCAR1_loglik Calculate REML Log-Likelihood of glsCAR1 model for each gene glsSymm Fit General Linear Model with corSymm Correlation Structure for One Gene myvoom myvoom function varbeta Recovering a Symmetric Matrix from Its Lower Triangular Matrix teststat Calculating F-Type Statistics To Test a General Linear Hypothesis my_splineDiffExprs Differential expression analysis based on natural cubic spline regression models for time-course data res Data Containing Results of Our Proposed Method Applying to RFI RNA-seq data jabes.q Q-value Using Histogram-based Method resSymm Data Containing results when analyzing RFI using TC_Symm shrink.phi Shrinkaged Estimates of Error Variance voomgls_CAR1 General Linear Model Using Voom Output sc_CAR1 Simulating Count Data From The Output of Real Data Analysis voomgls_Symm General Linear Model Using Voom Output corSymm correlction structure edgeRFit Analysis of the RFI RNA-seq data Using edgeR TC_CAR1 RNA-seq Analysis for Repeated-measures Data No Results!