evaluateDuplos: Evaluation of differences between duplicate samples

In rwehrens/BatchCorrMetabolomics: Complementary package to "Improved Batch Correction in Untargeted MS-Based Metabolomics"

Description

After batch correction, duplicate samples (either biological replicates or technical replicates) should be closer than in the uncorrected data. This function calculates the fraction of variance explained by the sample identity and the total variance, a measure also known as repeatability (or in genetic studies: heritability).

Usage

evaluateDuplos(X, Y, plot = !perMetabolite, perMetabolite = TRUE, ...)

Arguments

X	Data matrix.
Y	Data.frame describing information of the samples: columns SCode and Batch are mandatory.
plot	Logical: should a histogram of repeatabilities be shown?
perMetabolite	Separate repeatabilities for each metabolite (variable), or one average?
...	Further graphical arguments.

Value

The ratio of the amount of variation explained by the sample type (here: genotype) and the total variation. This is also called repeatability, or in a genetic context, heritability. Either given as one overall average, or per sample type.

Author(s)

Ron Wehrens

See Also

evaluatePCA

Examples

data(BC)
set.1.lod <- min(set.1[!is.na(set.1)])

## do correction for the first 10 metabolites
set.1.corrected.Q0 <-
  apply(set.1[,1:10], 2, doBC, ref.idx = which(set.1.Y$SCode == "ref"),
        batch.idx = set.1.Y$Batch, minBsamp = 4,
        seq.idx = set.1.Y$SeqNr, method = "lm",
        imputeVal = 0)
set.1.corrected.Q2 <-
  apply(set.1[,1:10], 2, doBC, ref.idx = which(set.1.Y$SCode == "ref"),
        batch.idx = set.1.Y$Batch, minBsamp = 4,
        seq.idx = set.1.Y$SeqNr, method = "lm",
        imputeVal = set.1.lod)

huhnDuplos.A0 <- evaluateDuplos(set.1.corrected.Q0, set.1.Y,
                             plot = TRUE, perMetabolite = FALSE)
title(main = paste("Q: repeatability:", round(huhnDuplos.A0, 3)),
      sub = "NA imputation: 0")
huhnDuplos.A2 <- evaluateDuplos(set.1.corrected.Q2, set.1.Y,
                             plot = TRUE, perMetabolite = FALSE)
title(main = paste("Q: repeatability:", round(huhnDuplos.A2, 3)),
      sub = "NA imputation: LOD")

## Per-metabolite information:
evaluateDuplos(set.1.corrected.Q2, set.1.Y, what = "Duplos",
               perMetabolite = TRUE)

rwehrens/BatchCorrMetabolomics documentation built on May 28, 2019, 10:42 a.m.