Set optimization boundaries for the model parameters.

Source: R/parameters.R

Set optimization boundaries for the model parameters.

setBoundaries(boundaries, normFactors = c(0.01, 10),
  options = .defaultParams)

Arguments

boundaries

a named list of lower and upper boundaries for the parameters which are used in the formulas.
normFactors

either a vector of length 2 or a list of two lists (lower, upper boundaries).
options

an options object to use as a basis for a new parameter set

Value

an options object with the new parameter values

Details

If no options object is provided, the default values are used. The elements in the boundaries list are either

  • a list of two vectors (the lower and upper boundaries for every gene/isoform respectively). If a vector is of length 1, equal boundary values will be assumed for all genes (e.g. for the lower boundary);

  • a vector of two scalars.

The normFactors elements (two, for the lower and the upper boundaries) can be:

  • a list with the structure is the same with the interSampleCoeffs in the PulseData object used in the analysis;

  • a list with the length equals the number of unique conditions; the structure is the same as formulaIndexes object used for the PulseData object creation;

  • a single scalar value.

Examples

# the simple way:
setBoundaries(list(a = c(1,2), b = c(10, 20)),
              normFactors = c(.1,10))
#> $tolerance
#> $tolerance$params
#> [1] 0.001
#> 
#> $tolerance$shared
#> [1] 0.01
#> 
#> $tolerance$normFactors
#> [1] 0.001
#> 
#> $tolerance$logLik
#> [1] 0.5
#> 
#> 
#> $verbose
#> [1] "silent"
#> 
#> $lb
#> $lb$size
#> [1] 10
#> 
#> $lb$a
#> [1] 1
#> 
#> $lb$b
#> [1] 10
#> 
#> $lb$normFactors
#> [1] 0.1
#> 
#> 
#> $ub
#> $ub$size
#> [1] 1e+07
#> 
#> $ub$a
#> [1] 2
#> 
#> $ub$b
#> [1] 20
#> 
#> $ub$normFactors
#> [1] 10
#> 
#> 
#> $fixedNorms
#> [1] FALSE
#> 

# the hard way:
# this are the formula indexes (see PulseData function documentation)
formulaIndexes <- list(
  total_fraction = 'total',
  pull_down      = c('labelled', 'unlabelled'),
  flow_through   = c('unlabelled', 'labelled')
)
# the lower and upper boundaries must have the same structure:
lbNormFactors <- list(
  total_fraction = 1,
  pull_down      = c(.1,.010),
  flow_through   = c(.1,.010))
ubNormFactors <- list(
  total_fraction = 1,
  pull_down      = c(10, 2),
  flow_through   = c(10, 2))
# we need to provide them as a list  
opts <- setBoundaries(
  list(mu = c(1, 1e6), d = c(1, 2)),
  normFactors = list(lbNormFactors, ubNormFactors))