md_predictions: Make model-based metadata predictions

In metamaden/rmpipeline: Process DNAm array IDATs into database files

Make model-based metadata predictions

Description

Make model-based metadata predictions using DNAm assays. Predictions for age, sex, and blood cell fractions are produced. Sex and blood cell predictions use raw/unnormalized red/green signal from h5se rg data (rgset.fname argument), while age predictions use noob-normalized Beta-values from h5se gr data (grset.fname argument).

Usage

md_predictions(
  ts,
  rgset.fname,
  grset.fname,
  mdmod.fname = "mdmod_dnam-predictions",
  files.dname = "recount-methylation-files",
  md.dname = "metadata",
  nsamp.block = 50,
  comp.dname = "compilations",
  verbose = TRUE
)

Arguments

ts	Timestamp for the preprocessed metadata table to output (integer or character).
rgset.fname	Name of the compilation file containing red signals extracted from IDATs (character).
grset.fname	Name of the compilation file containing green signals extracted from IDATs (character).
mdmod.fname	Name of the table of model-based predictions produced ("mdmod_dnam-predictions").
files.dname	Main recountmethylation instance files directory ("recount-methylation-files").
md.dname	Name of directory, in files.dname, containing the instance metadata files ("metadata).
nsamp.block	Number of samples per data block processed (integer, 50).
comp.dname	Name of directory, in files.dname, containing the compilation files with red and green signals for prediction calculations ("compilations")
verbose	Whether to show status messages (TRUE).
rgset.path	Path

Value

NULL, produces table of model-based metadata predictions.

See Also

get_qcmetrics(); md_preprocess(); md_postprocess()

metamaden/rmpipeline documentation built on Jan. 12, 2023, 3:20 p.m.