titan( data = NULL, trace = TRUE, widget = TRUE, dataFile = "", outFile = "", pdfFile = "", flagRaw = FALSE, flagFitted = FALSE, freqLo = .05, freqHi = .95, reg = "least.squares", term = "linear", sel = "wald", alpha = .05, rx0 = NULL, rx1 = NULL, gene0 = NULL, gene1 = NULL, R = 1000, seed = 0, ciConf = .95, ciType = "all"
)gene and treatment (or rx),
the numeric vectors frequency and competitor (or concentration),
and the numeric vector flag taking binary values. gene and treatment
specify the sample, and competitor the concentration of the competitor.
flag takes the value 0 for data points that are to be used in the analysis, and 1 for
data points that are to be ignored. If this parameter is set to NULL,
the user must supply the address of a data file containing this information:
see dataFile below.
TRUE, the output is printed as the function proceeds.
TRUE, the GUI is used to specify the data and set parameters for the
analysis. Otherwise, the following arguments must be supplied:
data parameter
above.
freqLo and freqHi.
freqLo and freqHi.
"least.squares",
"robust", or "spline" for least squares, robust, or natural spline regressions, respectively.
"linear", "quadratic" or "parallel.linear".
"linear" fits a regression with a linear term and intercept for each combination of gene and
treatment condition. "quadratic" also fits quadratic terms. "parallel.linear" estimates
only one linear term per gene, with intercepts for each combination of gene and treatment.
Natural spline regressions specify the number of degrees of freedom in the regression as either a
numeric or character vector.
"wald", indicating backwards selection by the Wald test, or "aic",
signifying stepwise selection by AIC. Robust regressions use only backwards selection by the Wald test.
Natural spline regressions use only stepwise selection by AIC.
treatment, against which
other treatments are compared when calculating fold changes. If NULL, the first level of treatment
is taken as the baseline condition.
treatment
that are entered into the analysis as test conditions when calculating fold changes.
If NULL, it is assumed that all the treatments other than that specified in rx0
are test conditions.
gene.
If NULL, it is assumed that none of the genes are housekeeping genes.
gene.
If NULL, it is assumed that all the genes other than those specified in gene0
are test genes.
c("norm","basic", "perc", "bca") or simply "all"
which will compute all four types of intervals.
"titan" containing the output.
it has components
flag
which contains the actual data flags used in the analysis.
lm or rlm, the output from the regression.
roots containing the interpolated values for each
combination of gene and treatment, i.e. the equivalence points.
If there is more than one treatment, the list contains a further matrix
log10fold containing the base 10 logarithms of the fold changes,
calculated relative to the baseline treatment, and adjusted for the fold changes
for the housekeeping or control genes.
boot, the output from the bootstrap.
titanbootci containing the boostrap confidence intervals.
data(hypoxia)
# Analyze titration series using GUI (code not run):
#
# t1 <- titan(data = hypoxia, R=0)
# Perform an identical reanalysis without using GUI:
#
# first get previous options
#
opt <- list(data = hypoxia, R=0)
#
# adjust the options so that
# no further data points are flagged
#
opt$flagRaw <- FALSE
opt$flagFitted <- FALSE
#
# turn off GUI
#
opt$widget <- FALSE
#
# call function
t2 <- do.call( "titan", opt )
# Reanalyze data with different parameters:
#
opt$reg <- "robust"
opt$sel <- "wald"
t3 <- do.call( "titan", opt )
Run the code above in your browser using DataLab