msNormalize: Function for performing normalization and batch corrections...

In KechrisLab/MSPrep: Package for Summarizing, Filtering, Imputing, and Normalizing Metabolomics Data

Description

Perform normalization and batch corrections on specified imputed dataset. Routines included are quantile, RUV (remove unwanted variation), SVA (surrogate variable analysis), median, CRMN (cross-contribution compensating multiple standard normalization), and ComBat to remove batch effects in raw, quantile, and median normalized data. Generates data driven controls if none exist.

Usage

msNormalize(
  data,
  normalizeMethod = c("median", "ComBat", "quantile", "quantile + ComBat",
    "median + ComBat", "CRMN", "RUV", "SVA"),
  nControl = 10,
  controls = NULL,
  nComp = 2,
  kRUV = 3,
  batch = "batch",
  covariatesOfInterest = NULL,
  transform = c("log10", "log2", "ln", "none"),
  compVars = c("mz", "rt"),
  sampleVars = c("subject_id"),
  colExtraText = NULL,
  separator = NULL,
  returnToSE = FALSE,
  returnToDF = FALSE
)

Arguments

data	Data set as either a data frame or 'SummarizedExperiement'.
normalizeMethod	Name of normalization method. "ComBat" (only ComBat batch correction), "quantile" (only quantile normalization), "quantile + ComBat" (quantile with ComBat batch correction), "median" (only median normalization), "median + ComBat" (median with ComBat batch correction), "CRMN" (cross-contribution compensating multiple standard normalization), "RUV" (remove unwanted variation), "SVA" (surrogate variable analysis)
nControl	Number of controls to estimate/utilize (for CRMN and RUV).
controls	Vector of control identifiers. Leave blank for data driven controls. Vector of column numbers from metafin dataset of that control (for CRMN and RUV).
nComp	Number of factors to use in CRMN algorithm.
kRUV	Number of factors to use in RUV algorithm.
batch	Name of the sample variable identifying batch.
covariatesOfInterest	Sample variables used as covariates in normalization algorithms (required for ComBat, CRMN, and SVA).
transform	Select transformation to apply to data prior to normalization. Options are "log10", "log2", "ln" and "none".
compVars	Vector of the columns which identify compounds. If a 'SummarizedExperiment' is used for 'data', row variables will be used.
sampleVars	Vector of the ordered sample variables found in each sample column.
colExtraText	Any extra text to ignore at the beginning of the sample columns names. Unused for 'SummarizedExperiments'.
separator	Character or text separating each sample variable in sample columns. Unused for 'SummarizedExperiment'.
returnToSE	Logical value indicating whether to return as 'SummarizedExperiment'
returnToDF	Logical value indicating whether to return as data frame.

Value

A data frame or 'SummarizedExperiment' with transformed and normalized data. Default return type is set to match the data input but may be altered with the 'returnToSE' or 'returnToDF' arguments.

References

Bolstad, B.M.et al.(2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics, 19, 185-193

DeLivera, A.M.et al.(2012) Normalizing and Integrating Metabolomic Data. Anal. Chem, 84, 10768-10776.

Gagnon-Bartsh, J.A.et al.(2012) Using control genes to correct for unwanted variation in microarray data. Biostatistics, 13, 539-552.

Johnson, W.E.et al.(2007) Adjusting batch effects in microarray expression data using Empirical Bayes methods. Biostatistics, 8, 118-127.

Leek, J.T.et al.(2007) Capturing Heterogeneity in Gene Expression Studies by Surrogate Variable Analysis. PLoS Genetics, 3(9), e161

Wang, W.et al.(2003) Quantification of Proteins and Metabolites by Mass Spectrometry without Isotopic Labeling or Spiked Standards. Anal. Chem., 75, 4818-4826.

Examples

# Load, tidy, summarize, filter, and impute example dataset
data(msquant)

summarizedDF <- msSummarize(msquant,
                            compVars = c("mz", "rt"),
                            sampleVars = c("spike", "batch", "replicate", 
                            "subject_id"),
                            cvMax = 0.50,
                            minPropPresent = 1/3,
                            colExtraText = "Neutral_Operator_Dif_Pos_",
                            separator = "_",
                            missingValue = 1)
                            
filteredDF <- msFilter(summarizedDF,
                       filterPercent = 0.8,
                       compVars = c("mz", "rt"),
                       sampleVars = c("spike", "batch", "subject_id"),
                       separator = "_")

hmImputedDF <- msImpute(filteredDF, imputeMethod = "halfmin",
                        compVars = c("mz", "rt"),
                        sampleVars = c("spike", "batch", "subject_id"),
                        separator = "_",
                        missingValue = 0)

# Normalize data set
medianNormalizedDF <- msNormalize(hmImputedDF, normalizeMethod = "median",
                                  compVars = c("mz", "rt"),
                                  sampleVars = c("spike", "batch", 
                                  "subject_id"),
                                  separator = "_")

KechrisLab/MSPrep documentation built on Feb. 2, 2022, 2:43 a.m.