Evaluate performance of DC methods on simulations

Quantify the performance of a differential co-expression pipeline on simulated data.

dcEvaluate(
  simulation,
  dclist,
  truth.type = c("association", "influence", "direct"),
  perf.method = "f.measure",
  combine = TRUE,
  ...
)

Arguments

simulation

a list, storing data and results generated from simulations

dclist

a list of igraphs, produced using dcPipeline

truth.type

a character, specifying which level of the true network to retrieve: 'association' (default), 'influence' or 'direct'

perf.method

a character, specifying the method to use. Available methods can be accessed using perfMethods

combine

a logical, indicating whether differential networks from independent knock-outs should be treated as a single inference or independent inferences (defaults to TRUE)

...

additional parameters to be passed on to the performance metric method (see performanceMeasure)

Value

a numeric, representing the performance metric. A single value if

combine = TRUE and a named vector otherwise.

See also

dcPipeline, performanceMeasure, perfMethods

Examples

data(sim102)

#run a standard pipeline
resStd <- dcPipeline(sim102, dc.func = 'zscore')
dcEvaluate(sim102, resStd)
#> [1] 0.5625966
dcEvaluate(sim102, resStd, combine = FALSE)
#>      ADR1      UME6 
#> 0.7228916 0.3936508