trendy: Trendy

In rhondabacher/Trendy: Breakpoint analysis of time-course expression data

Trendy

Description

Segmented regression models are fit for each gene. The number of model fits is 1 -> maxK.

Usage

trendy(Data = NULL, tVectIn = NULL, saveObject = FALSE,
  fileName = NULL, meanCut = 10, maxK = 3, minNumInSeg = 5,
  pvalCut = 0.1, numTry = 5, keepFit = FALSE, NCores = NULL,
  featureNames = NULL)

Arguments

Data	a matrix of normalized expression measurements. Rows are genes/features and columns are samples.
tVectIn	a numerical vector indicating the time-points or the order of samples. If it is NULL (default), then the time/order will be assumed to be equaly spaced from 1:N (N is number of samples).
saveObject	if TRUE then the trendy object produced will be saved to use in the Shiny app (default is FALSE).
fileName	the file name (and file path) to save the Trendy object, only used if saveObject=TRUE (default name is trendyOutputForShiny.RData).
meanCut	genes whose mean is less than MeanCut will not be considered, default is 10.
maxK	maximum number of breakpoints to consider. For each gene, trendy will fit maxK + 1 models containing 0 -> maxK breakpoints (1 -> (maxK + 1) segments). The model with the lowest BIC value will be selected (unless forceRsq = TRUE, see below).
minNumInSeg	minimum number of samples required to be within a segment. If a breakpoint model has a segment with fewer than minNumInSeg point in any segment, then the model is not considered valid.
pvalCut	p-value cutoff. If the p-value of a segment is greater than PvalCut, then the segment will be called as 'no change'.
numTry	the number of different seeds to try. If all numTry runs fail, then the linear regression (no breakpoints, one segment) model will be returned.
keepFit	whether to report the fitted object (default is FALSE).
NCores	number of cores to use, default is detectCores() - 1.
featureNames	optional parameter to specify an explicit subset of features/genes to fit the segmented regression model to.

Value

Trend: direction of each sample; -1: down, 0: no change, 1: up Slope: fitted slopes, Slope.Trend: sign of fitted slopes, Slope.Pvalue: p value of each segment, Breakpoint: estimated breakpoints, Fitted.Values: fitted values AdjustedR2: adjusted R squared value of the model Fit: fit object

Author(s)

Ning Leng and Rhonda Bacher

Examples

 m1 <- matrix(c(c(rnorm(50,5,1),sort(rnorm(50, 15, 5))), rnorm(100, 50,10)), 2, 100, TRUE)
 rownames(m1) <- c("g1","g2")
 colnames(m1) <- paste0("time", seq_len(100))
 myTrends <- trendy(m1)

rhondabacher/Trendy documentation built on Oct. 26, 2023, 3:46 a.m.