metagenomeSeq (version 1.14.0)

fitSSTimeSeries: Discover differentially abundant time intervals using SS-Anova

Description

Calculate time intervals of interest using SS-Anova fitted models. Fitting is performed uses Smoothing Spline ANOVA (SS-Anova) to find interesting intervals of time. Given observations at different time points for two groups, fitSSTimeSeries calculates a function that models the difference in abundance between two groups across all time. Using permutations we estimate a null distribution of areas for the time intervals of interest and report significant intervals of time. Use of the function for analyses should cite: "Finding regions of interest in high throughput genomics data using smoothing splines" Talukder H, Paulson JN, Bravo HC. (In preparation)

Usage

fitSSTimeSeries(obj, formula, feature, class, time, id, lvl = NULL, include = c("class", "time:class"), C = 0, B = 1000, norm = TRUE, log = TRUE, sl = 1000, ...)

Arguments

obj

metagenomeSeq MRexperiment-class object.

formula

Formula for ssanova.

feature

Name or row of feature of interest.

class

Name of column in phenoData of MRexperiment-class object for class memberhip.

time

Name of column in phenoData of MRexperiment-class object for relative time.

id

Name of column in phenoData of MRexperiment-class object for sample id.

lvl

Vector or name of column in featureData of MRexperiment-class object for aggregating counts (if not OTU level).

include

Parameters to include in prediction.

C

Value for which difference function has to be larger or smaller than (default 0).

B

Number of permutations to perform

norm

When aggregating counts to normalize or not.

log

Log2 transform.

sl

Scaling value.

...

Options for ssanova

Value

List of matrix of time point intervals of interest, Difference in abundance area and p-value, fit, area permutations, and call.A list of objects including:

• timeIntervals - Matrix of time point intervals of interest, area of differential abundance, and pvalue.
• data - Data frame of abundance, class indicator, time, and id input.
• fit - Data frame of fitted values of the difference in abundance, standard error estimates and timepoints interpolated over.
• perm - Differential abundance area estimates for each permutation.
• call - Function call.

See Also

cumNorm ssFit ssIntervalCandidate ssPerm ssPermAnalysis plotTimeSeries

Examples

data(mouseData)
res = fitSSTimeSeries(obj=mouseData,feature="Actinobacteria",
   class="status",id="mouseID",time="relativeTime",lvl='class',B=2)