fit-methods: Fit model on time series data

In dipenps/rnits: R Normalization and Inference of Time Series data

Fit model on time series data

Description

Fit a model comparing time series data set Rnits objects

Usage

fit(object, cluster = TRUE, B = 100, verbatim = FALSE, nclus = NULL,
  modelhistplot = FALSE, seed = 123, gene.level = TRUE,
  clusterallsamples = FALSE, model = NULL)

## S4 method for signature 'Rnits'
fit(object, cluster = TRUE, B = 100, verbatim = FALSE,
  nclus = NULL, modelhistplot = FALSE, seed = 123, gene.level = TRUE,
  clusterallsamples = FALSE, model = NULL)

Arguments

object	Rnits object
cluster	if TRUE, perform clustering to identify groups of genes/probes with similar expression profiles.
B	Default 100. Number of bootstrap iterations for p-value calculation
verbatim	If FALSE, print out details of fitting models.
nclus	Default NULL. Number of clusters to use for k-means clustering.
modelhistplot	If TRUE, p-value histograms of multiple models are plotted.
seed	Random seed for bootstrap iterations
gene.level	If TRUE, collapse probes to gene level information.
clusterallsamples	If TRUE, Use all time series for clustering. By default, only the sample labeled 'control' is used or the lexically first sample is used.
model	A data frame with fields 'degree' and 'df' indicating a specific B-spline model to be used. If provided, model selection is not run.

Details

The function compares multiple time-series expression data sets by i) (optional) summarizing probes into gene-level information ii) (optional) identifying a set of co-expressed genes by clustering iii) For each cluster (or for all genes /probes), fit a series of B-splines with varying curvature and degrees of freedom. Under the null hypothesis H_0, a single model is fit for all data sets, while under H_1, each data set is fit separately. P-values from the hypothesis test are then plotted and the least complex spline parameters that result in uniformly distributed null p-values are automatically chosen.

Value

An object of S4 class Rnits with fitted results data containing cluster information, ratio statistics and p-values.

Examples

# load pre-compiled expressionSet object for Ronen and Botstein yeast  chemostat data
data(yeastchemostat)
rnitsobj = build.Rnits(yeastchemostat, logscale = TRUE, normmethod = 'Between')
## Not run: 
# Fit model using gene-level summarization
rnitsobj <- fit(rnitsobj, gene.level = TRUE, clusterallsamples = FALSE)

## End(Not run)

dipenps/rnits documentation built on March 18, 2023, 7:30 p.m.