rtpcr (version 2.1.4)

ANOVA_DDCt: Delta Delta Ct ANOVA analysis with optional model specification

Description

Apply Delta Delta Ct (ddCt) analysis to each target gene and performs per-gene statistical analysis.

Usage

ANOVA_DDCt(
  x,
  numOfFactors,
  numberOfrefGenes,
  mainFactor.column,
  block,
  mainFactor.level.order = NULL,
  p.adj = "none",
  analyseAllTarget = TRUE,
  model = NULL,
  set_missing_target_Ct_to_40 = FALSE,
  se.type = c("paired.group", "two.group", "single.group"),
  modelBased_se = TRUE
)

Value

An object containing expression table, lm model, residuals, raw data and ANOVA table for each gene:

ddCt expression table along with per-gene statistical comparison outputs

object$relativeExpression

ANOVA table

object$perGene$gene_name$ANOVA_table

lm ANOVA

object$perGene$gene_name$lm

lm_formula

object$perGene$gene_name$lm_formula

Residuals

resid(object$perGene$gene_name$lm)

Arguments

x

The input data frame containing experimental design columns, replicates (integer), target gene E/Ct column pairs, and reference gene E/Ct column pairs. Reference gene columns must be located at the right end of the data frame. See "Input data structure" in vignettes for details about data structure.

numOfFactors

Integer. Number of experimental factor columns (excluding rep and optional block).

numberOfrefGenes

Integer. Number of reference genes.

mainFactor.column

Integer. Column index of the factor for which the relative expression analysis is applied.

block

Character. Block column name or NULL. When a qPCR experiment is done in multiple qPCR plates, variation resulting from the plates may interfere with the actual amount of gene expression. One solution is to conduct each plate as a randomized block so that at least one replicate of each treatment and control is present on a plate. Block effect is usually considered as random and its interaction with any main effect is not considered.

mainFactor.level.order

Optional character vector specifying the order of levels for the main factor. If NULL, the first observed level is used as the calibrator. If provided, the first element of the vector is used as the calibrator level.

p.adj

Method for p-value adjustment. See p.adjust.

analyseAllTarget

Logical or character. If TRUE (default), all target genes are analysed. Alternatively, a character vector specifying the names (names of their Efficiency columns) of target genes to be analysed.

model

Optional model formula. If provided, this overrides the automatic formula (uni - or multi-factorial CRD or RCBD based on block and numOfFactors). The formula uses wDCt as the response variable. For mixed models, random effects can be defined using lmer syntax (e.g., "wDCt ~ Treatment + (1 | id)"). When using model, the block and numOfFactors arguments are ignored for model specification, but still used for data structure identification.

for fixed effects only, the "lm" (ordinary least squares) is used. "lmer" is used for mixed effects models (requires the lmerTest package). If a custom formula is provided with random effects, the function will use lmerTest::lmer(); otherwise it will use stats::lm(). Note that emmeans supports both model types and will use appropriate degrees of freedom methods (Satterthwaite by default).

set_missing_target_Ct_to_40

If TRUE, missing target gene Ct values become 40; if FALSE (default), they become NA.

se.type

Character string specifying how standard error is calculated. One of "paired.group", "two.group", or "single.group". "paired.group" computes SE from paired differences (used when a random id effect is present), "two.group" uses the unpaired two-group t-test standard error against the reference level, and "single.group" computes SE within each level using a one-group t-test.

modelBased_se

Logical. If TRUE (default), standard errors are calculated from model-based residuals. If FALSE, standard errors are calculated directly from the observed wDCt values within each treatment group according to the selected se.type. For single factor data, both methods are the same. It is recommended to use modelBased_se = TRUE (default).

Details

ddCt analysis of variance (ANOVA) is performed for the mainFactor.column based on a full model factorial experiment by default. However, if ANCOVA_DDCt function is used, analysis of covariance is performed for the levels of the mainFactor.column and the other factors are treated as covariates. if the interaction between the main factor and the covariate is significant, ANCOVA is not appropriate.

All the functions for relative expression analysis (including TTEST_DDCt(), WILCOX_DDCt(), ANOVA_DDCt(), and ANOVA_DCt()) return the relative expression table which include fold change and corresponding statistics. The output of ANOVA_DDCt(), and ANOVA_DCt() also include lm models, residuals, raw data and ANOVA table for each gene.

The expression table returned by TTEST_DDCt(), WILCOX_DDCt(), and ANOVA_DDCt() functions include these columns: gene (name of target genes), contrast (calibrator level and contrasts for which the relative expression is computed), ddCt (mean of weighted delta delta Ct values), RE (relative expression or fold change = 2^-ddCt), log2FC (log(2) of relative expression or fold change), pvalue, sig (per-gene significance), LCL (95% lower confidence level), UCL (95% upper confidence level), se (standard error of mean calculated from the weighted delta Ct values of each of the main factor levels), Lower.se.RE (The lower limit error bar for RE which is 2^(log2(RE) - se)), Upper.se.RE (The upper limit error bar for RE which is 2^(log2(RE) + se)), Lower.se.log2FC (The lower limit error bar for log2 RE), and Upper.se.log2FC (The upper limit error bar for log2 RE)

References

LivakKJ, Schmittgen TD (2001). Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the Double Delta CT Method. Methods, 25(4), 402–408. doi:10.1006/meth.2001.1262

Ganger MT, Dietz GD, and Ewing SJ (2017). A common base method for analysis of qPCR data and the application of simple blocking in qPCR experiments. BMC Bioinformatics, 18, 1–11.

Taylor SC, Nadeau K, Abbasi M, Lachance C, Nguyen M, Fenrich, J. (2019). The ultimate qPCR experiment: producing publication quality, reproducible data the first time. Trends in Biotechnology, 37, 761-774.

Yuan JS, Reed A, Chen F, Stewart N (2006). Statistical Analysis of Real-Time PCR Data. BMC Bioinformatics, 7, 85.

Examples

data1 <- read.csv(system.file("extdata", "data_2factorBlock3ref.csv", package = "rtpcr"))
ANOVA_DDCt(data1,
           numOfFactors = 2,
           numberOfrefGenes = 3,
           block = "block",
           mainFactor.column = 2,
           p.adj = "none")
           
data2 <- read.csv(system.file("extdata", "data_1factor_one_ref.csv", package = "rtpcr"))          
ANOVA_DDCt(data2,
           numOfFactors = 1,
           numberOfrefGenes = 1,
           block = NULL,
           mainFactor.column = 1,
           p.adj = "none",
           se.type = "single.group")
  
# # Repeated measure analysis         
# a <- ANOVA_DDCt(data_repeated_measure_1,
#            numOfFactors = 1,
#            numberOfrefGenes = 1,
#            block = NULL,
#            mainFactor.column = 1,
#            p.adj = "none", model = wDCt ~ time + (1 | id))
# 
# a$perGene$Target$ANOVA_table
# 
# 
# # Repeated measure analysis: split-plot in time
# a <- ANOVA_DDCt(data_repeated_measure_2,
#            numOfFactors = 2, numberOfrefGenes = 1,
#            mainFactor.column = 2, block = NULL,
#            model = wDCt ~ treatment * time + (1 | id))