QC_methylation: Quality control of methylation array data

In frankRuehle/systemsbio: Streamlined Analysis and Integration of Systems Biology Data

Description

Multiple QC procedures of an RGChannelSet object containing methylation data.

Usage

QC_methylation(
  RGset,
  projectfolder = "MT/QC",
  projectname = NULL,
  sampleColumn = "Sample_Name",
  groupColumn = "Sample_Group",
  preprocessing = c("raw", "illumina", "swan"),
  QC_procedures = c("minfi", "WGCNA"),
  numPositionsMDSplot = 1000,
  methPlatform = "450k-UCSC",
  phDendro = "Sample_Group",
  flashClustMethod = "average",
  cex.dendroLabels = 0.6,
  threshold.Z.k = -5
)

Arguments

RGset	RGChannelSet object with methylation data.
projectfolder	character with directory for output files (will be generated if not exisiting).
projectname	optional character prefix for output file names.
sampleColumn	character with column name of Sample names in RGset.
groupColumn	character with column name of group names in RGset.
preprocessing	character with preprocessing procedures to apply. Any of "raw", "illumina", "swan". If "swan" is selected, at least one more procedure must be selected to.
QC_procedures	character with desired QC procedures to use. Any of "minfi", "COHCAP", "WGCNA".
numPositionsMDSplot	number of genomic positions with the most methylation variability to be used for calculating distance between samples in MDS plots (needed for minfi only).
methPlatform	Annotation file to be used. Enter "450k-UCSC" for UCSC CpG Islands for 450k array probes, " 450k-HMM" for HMM CpG Islands for 450k array probes, "27k" for UCSC CpG Islands for 27k array probes (needed for COHCAP only).
phDendro	character vector with phenotypes of RGset to be displayed in sample dendrogram (needed for WGCNA only).
flashClustMethod	character vector with the agglomeration method to be used in WGCNA package. Can be one of "ward", "single", "complete", "average", "mcquitty", "median" or "centroid" (needed for WGCNA only).
cex.dendroLabels	numeric. Controls size of labels sample in dendrogram (needed for WGCNA only).
threshold.Z.k	numeric. Threshold for outlier prediction in WGCNA package. Outlier samples are highlighted in red in sample dendrogram (needed for WGCNA only).

Details

This function takes an RGChannelSet as input and transforms it to a MethylSet object using 3 preprocessing procedures from minfi package:

• preprocessRaw (no normalisiation)

• preprocessIllumina applies background subtraction as well as control normalization like in Illumina Genome studio

• preprocessSWAN applies SWAN normalization method on a preprocessed MethylSet to reduce the technical variability of Type I and Type II Illumina probes. For this, Illumina preprocessed MethylSet is used if given, the raw preprocessed MethylSet otherwise.

Quality control functions are applied to all 3 MethylSet objects. The required annotation packages (currently hg19 only) IlluminaHumanMethylation450kmanifest and IlluminaHumanMethylation450kanno.ilmn12.hg19 are installed and/or loaded automatically. Required packages not already attached when running QC_methylation are detached afterwards. The function generates the following plots using quality control functionalities from up to 3 packages:

• minfi

  • qcReport using array control probes (for RGChannelSet)

  • Median intensity plots

  • Predicted sex plots

  • Multi-dimensional scaling (MDS) plots

  • Histogram overview of Illumina and SWAN normalisation

• COHCAP (needs much computational resources)

  • Cluster plot

  • PCA plot

  • histogram plot

• WGCNA

  • clustered sample dendrogram

  • indicated outlier detection

Value

no value returned. QC plots are stored as side-effects.

Author(s)

Frank Ruehle

frankRuehle/systemsbio documentation built on Sept. 14, 2020, 1:18 a.m.