cq2counts {MCMC.qpcr}

Prepares qRT-PCR data for mcmc.qpcr analysis

Description

Converts Cq values into molecule counts, and stacks the dataset

Usage

cq2counts(data, genecols, condcols, effic, Cq1 = NULL)

Arguments

data

Raw qRT-PCR dataset, one Cq column per gene, plus columns containing factors. The Cq columns, in addition to the proper Cq values, may contain NA (missing data) and -1, which means no amplification observed (i.e., zero target molecules at the start of qPCR reaction). Column headers are either gene names or factor names. Any number of fixed factors is allowed; any number of random factors that are gene-specific scalars (such as effect of genotype, or block) Must have a column called "sample", denoting individual cDNA preps. Technical replicates should not be averaged, they should be represented as independent rows with the same sample ID.

genecols

columns that contain Cq data

condcols

columns corresponding to factors, including "sample" factor

effic

The PCR efficiency data for each of the analyzed genes. This is data frame with two columns: gene name (must exactly match the headers of gene columns in Cq data table!) and efficiency (fold- amplification per PCR cycle, determined from qPCR of serial dilutions; see PrimEff() function )

Cq1

The Cq of a single molecule. If left unspecified, it will be calculated from the efficiency (E) using approximate formula Cq1=51.6-7.56*E, derived empirically for Roche's LightCycler 480. Cq1 does not seem to have much effect on relative quantification results unless it is wildly off (by 2-3 cycles). For an unknown qPCR instrument a single Cq1=37 could be assumed for all genes.

Note: If all experimental Cq values are less than 30, Cq1 variation (within a reasonable range, 35-39) will not have any effect on the results whatsoever, so just go for Cq1=37.

Value

Returns a dataframe with a single response variable column ("count"), gene column ("gene") and several columns containing factors.

Note: The purpose of Cq to counts conversion is to enable generalized linear model analysis, which would take care of the heteroscedasticity and occasional 'empty' amplification trials for low-abundant targets. Although this works well, the absolute values of molecule counts returned by cq2counts are still approximate, so they should not be relied upon for true single-molecule analysis.

Author(s)

Mikhail V. Matz, UT Austin <[e-mail]>

References

Matz MV, Wright RM, Scott JG (2013) No Control Genes Required: Bayesian Analysis of qRT-PCR Data. PLoS ONE 8(8): e71448. doi:10.1371/journal.pone.0071448

Examples

data(coral.stress) 
data(amp.eff) 
genecolumns=c(5:19) # specifying where the Ct data are in the data table
conditions=c(1:4) # specifying data table columns containing factors  
# calculating molecule counts and reformatting:
dd=cq2counts(data=coral.stress,genecols=genecolumns,
condcols=conditions,effic=amp.eff,Cq1=37)