sigma_control-class: Control parameters for seaMass-Σ

In biospi/deamass: Statistical modelling for mass spectrometry omics

Control parameters for seaMass-Σ

Description

Define control parameters for the seaMass-Σ Bayesian model.

Usage

sigma_control(
  keep = c("summaries", "group.quants"),
  summarise = c("groups", "components", "measurements"),
  plot = c("assay.stdevs", "group.means", "group.quants", "group.quants.pca",
    "component.means", "component.stdevs", "component.deviations",
    "component.deviations.pca", "measurement.means", "measurement.stdevs"),
  eb.model = "deconvolve",
  eb.max = 1024,
  measurement.model = "independent",
  measurement.eb.min = 2,
  component.model = "independent",
  component.eb.min = 3,
  assay.model = "component",
  assay.eb.min = 3,
  assay.eb.nsample = 16,
  error.model = "lognormal",
  missingness.model = "censored",
  missingness.threshold = 0,
  nchain = 4,
  nwarmup = 256,
  thin = 4,
  nsample = 1024,
  random.seed = 0,
  nthread = parallel::detectCores()%/%2,
  schedule = schedule_local()
)

Arguments

keep	Outputs to keep, NULL or a subset of c("summaries", "model0", "markdown", "assay.means", "group.quants", "group.means", "component.deviations", "component.means", "component.stdevs", "measurement.means", "measurement.stdevs")
summarise	Outputs to write csv summaries for, NULL or a subset of c("groups", "components", "measurements") Note, you must summarise or keep "standardised.group.deviations" if you want to run seaMass-Δ!
plot	Outputs to plot, NULL or a subset of c("assay.stdevs", "group.means", "group.quants", "group.quants.pca", "component.means", "component.stdevs", "component.deviations", "component.deviations.pca", "measurement.means", "measurement.stdevs")
eb.model	Empirical Bayes model, either NULL (none), "fit" (inverse Nakagami distribution fit) or "deconvolve" (LIMMA style deconvolution; default)
eb.max	Maximum number of components and measurements to use in empirical Bayes models.
measurement.model	Either "single" (single residual) or "independent" (per-measurement independent residuals; default)
measurement.eb.min	Minimum number of measurements per component to use for computing Empirical Bayes priors
component.model	Either NULL (no component model), "single" (single random effect) or "independent" (per-component independent random effects; default)
component.eb.min	Minimum number of components per group to use for computing Empirical Bayes priors
assay.model	Either NULL (no assay model), "measurement" (per-assay independent random effects across measurements) or "componenet" (per-assay independent random effects across components; default)
assay.eb.min	Minimum number of assays per group group to use for computing empirical Bayes priors
assay.eb.nsample	Number of MCMC samples to use for assay model input
error.model	Likelihood model, either "lognormal" (default) or "poisson"
missingness.model	Either NULL (do nothing), "rm" (NAs removed), "one" (NAs set to 1), "minimum" (NAs set to lowest quant of that measurement) or "censored" (NAs modelled as censored below lowest quant of that measurement; default)
missingness.threshold	All datapoints equal to or below this count are treated as missing
nchain	Number of MCMC chains to run
nwarmup	Number of MCMC warmup iterations to run for each chain
thin	MCMC thinning factor
nsample	Total number of MCMC samples to deliver downstream
random.seed	Random number seed
schedule	Either schedule_local (execute locally), schedule_pbs or schedule_slurm (prepare for submission to HPC cluster)

Functions

• sigma_control: Generator function

biospi/deamass documentation built on May 20, 2023, 3:30 a.m.