tsJumpStats: Compute statistics on jumps through community space

In hallucigenia-sparsa/seqtime: Time Series Analysis of Sequencing Data

Compute statistics on jumps through community space

Description

Consecutive samples representing time points in an ordination plot can be interpreted as vectors, which have a length and an angle. If ordinate is true, these lengths and angles are computed from the selection dimensions of the ordination. If ordinate is false, dissimilarities between consecutive time points are computed. If a perturbation object is provided and plot.type hist or box is true, the histogram or box plot of the perturbed and non-perturbed jump lengths is plotted and the significance of the difference between perturbed and non-perturbed jump lengths is assessed with a Wilcoxon test. Note that jump lengths plots are a visualization of beta-diversity, which is however not computed between all pair-wise but only between consecutive samples.

Usage

tsJumpStats(x, time.given = FALSE, time.unit = "days",
  ordinate = TRUE, distance = "bray", dimensions = c(1, 2),
  groups = c(), min.jump.length = 0, max.jump.length = Inf,
  plot = FALSE, plot.type = "bar", header = "", subsample = FALSE,
  computeCentroid = FALSE, perturb = NULL, verbose = FALSE)

Arguments

x	a community time series, with rows as taxa and columns as time points
time.given	sample names provide time steps, only needed for plotting and plot.type jumps
time.unit	unit of time, only needed for plotting and plot.type jumps
ordinate	if TRUE, compute jumps in ordination plot
distance	the distance (or dissimilarity) used to compute jumps directly or to compute the ordination
dimensions	if ordinate is TRUE, the principal components considered
groups	a vector with group assignments and as many entries as there are samples
min.jump.length	minimum length of jumps to be considered for the statistic
max.jump.length	maximum length of jumps to be considered for the statistic
plot	if TRUE, a plot of the jumps is made
plot.type	bar: plot jumps as bars in the order of occurrence; box: a box plot of jump lengths; hist: a histogram of jump lengths, power: power law of jumps of a certain length, msd: mean square displacement vs time lag
header	text to be used as plot title instead of default text
subsample	for plot.type hist or box: subsample larger jump vector randomly down to smaller one, so that there are as many perturbed as non-perturbed jumps
computeCentroid	whether or not to formally compute the centroid, if false, centroid is set to 0,0 (only for ordinate=TRUE)
perturb	a perturbation object, if provided, the plot is colored accordingly
verbose	depending on configuration, report details on jump statistics and Wilcoxon test result, values for thresholds, probabilities of transitions and position of centroid

Value

jump lengths (i.e. dissimilarities of community composition at consecutive time points) and, in case ordinate is true, angles are returned

Examples

## Not run: 
data(david_stoolA_otus)
data(david_stoolA_metadata)
rarefaction=rarefyFilter(david_stoolA_otus,min = 10000)
rarefiedA=rarefaction$rar
days=david_stoolA_metadata[1,rarefaction$colindices]
interpA=interpolate(rarefiedA,time.vector=days,interval=1,method="stineman")
interpA[interpA<0]=0
perturbA=perturbation(times=c(80,105), durations=c(5,9))
res=tsJumpStats(interpA, plot=TRUE, perturb=perturbA, header="in stool A")
out=powerspec(res$lengths,plot=TRUE)

## End(Not run)

hallucigenia-sparsa/seqtime documentation built on Jan. 9, 2023, 11:53 p.m.