melonnpan.train: Model-based Genomically Informed High-dimensional Predictor...

In biobakery/melonnpan: Model-based Genomically Informed High-dimensional Predictor of Microbial Community Metabolite Profiles

Model-based Genomically Informed High-dimensional Predictor of Microbial Community Metabolite Profiles

Description

Predict metabolites from microbial sequence features' abundances. Both measurements are expected to be normalized before using MelonnPan.

Usage

melonnpan.train(
  metab,
  metag,
  output,
  alpha = seq(0.05, 0.95, 0.05),
  lambda.choice = "lambda.1se",
  nfolds = 10,
  correction = "fdr",
  method = "spearman",
  cores = 4,
  seed = 1234,
  cutoff = 0.05,
  verbose = TRUE,
  no.transform.metab = FALSE,
  no.transform.metag = FALSE,
  discard.poor.predictions = FALSE,
  plot = FALSE,
  outputString = c("MelonnPan_Training_Summary", "MelonnPan_Trained_Weights",
    "MelonnPan_Trained_Metabolites")
)

Arguments

metab	Training data of metabolite relative abundances (normalized). Must have the exact same rows (subjects/samples) as metag.
metag	Training data of metagenomics sequence features' relative abundances (normalized). Must have the exact same rows (subjects/samples) as metab.
output	The output folder to write results.
alpha	Grid of alpha values between 0 and 1. Default is 'seq(0.05, 0.95, 0.05)'.
lambda.choice	Choice of optimal lambda ('lambda.min' or 'lambda.1se'). Default is 'lambda.1se'.
nfolds	Number of folds for internal cross-validation. Default is 10.
correction	Multiplicity adjustment method, same as 'p.adjust'. Default is 'fdr'.
method	Method to correlate measured and predicted metabolites ('spearman' or 'pearson'). Default is 'spearman'.
cores	Number of cores to use for parallel processing. Default is 4.
seed	Specify the arbitrary seed value for reproducibility. Default is 1234.
cutoff	Q-value threshold for significant prediction. Default is 0.05.
verbose	Should detalied message be printed. Default is TRUE.
no.transform.metab	Should arcsine square root transformation (AST) be turned off for 'metab'? Default is FALSE. If FALSE, 'metab' must be proportional data ranging from 0 to 1.
no.transform.metag	Should rank-based inverse normal (RIN) transformation be turned off for 'metag'? Default is FALSE.
discard.poor.predictions	Should predictions with model size = 1 be discarded? Default is FALSE.
plot	Should CV error as a function of lambda be plotted. Default is FALSE.
outputString	Names of the three output files. Default is 'MelonnPan_Training_Summary', 'MelonnPan_Trained_Weights', and 'MelonnPan_Trained_Metabolites'.

biobakery/melonnpan documentation built on March 26, 2024, 11:42 p.m.