hierarchicalConsensusCalculation {WGCNA}

Hierarchical consensus calculation

Description

Hierarchical consensus calculation with optional data calibration.

Usage

hierarchicalConsensusCalculation(
   individualData,

   consensusTree,

   level = 1,
   useBlocks = NULL,
   randomSeed = NULL,
   saveCalibratedIndividualData = FALSE,
   calibratedIndividualDataFilePattern = 
         "calibratedIndividualData-%a-Set%s-Block%b.RData",

   # Return options: the data can be either saved or returned but not both.
   saveConsensusData = TRUE,
   consensusDataFileNames = "consensusData-%a-Block%b.RData",
   getCalibrationSamples= FALSE,

   # Return the intermediate results as well?
   keepIntermediateResults = FALSE,

   # Internal handling of data
   useDiskCache = NULL, 
   chunkSize = NULL,
   cacheDir = ".",
   cacheBase = ".blockConsModsCache",

   # Behaviour
   collectGarbage = FALSE,
   verbose = 1, indent = 0)

Arguments

individualData

Individual data from which the consensus is to be calculated. It can be either a list or a multiData structure. Each element in individulData can in turn either be a numeric object (vector, matrix or array) or a BlockwiseData structure.

consensusTree

A list specifying the consensus calculation. See details.

level

Integer which the user should leave at 1. This serves to keep default set names unique.

useBlocks

When individualData contains BlockwiseData, this argument can be an integer vector with indices of blocks for which the calculation should be performed.

randomSeed

If non-NULL, the function will save the current state of the random generator, set the given seed, and restore the random seed to its original state upon exit. If NULL, the seed is not set nor is it restored on exit.

saveCalibratedIndividualData

Logical: should calibrated individual data be saved?

calibratedIndividualDataFilePattern

Pattern from which file names for saving calibrated individual data are determined. The conversions %a, %s and %b will be replaced by analysis name, set number and block number, respectively.

saveConsensusData

Logical: should final consensus be saved (TRUE) or returned in the return value (FALSE)?

consensusDataFileNames

Pattern from which file names for saving the final consensus are determined. The conversions %a and %b will be replaced by analysis name and block number, respectively.

getCalibrationSamples

When calibration method in the consensusOptions component of ConsensusTree is "single quantile", this logical argument determines whether the calibration samples should be returned within the return value.

keepIntermediateResults

Logical: should results of intermediate consensus calculations (if any) be kept? These are always returned as BlockwiseData whose data are saved to disk.

useDiskCache

Logical: should disk cache be used for consensus calculations? The disk cache can be used to store chunks of calibrated data that are small enough to fit one chunk from each set into memory (blocks may be small enough to fit one block of one set into memory, but not small enough to fit one block from all sets in a consensus calculation into memory at the same time). Using disk cache is slower but lessens the memory footprint of the calculation. As a general guide, if individual data are split into blocks, we recommend setting this argument to TRUE. If this argument is NULL, the function will decide whether to use disk cache based on the number of sets and block sizes.

chunkSize

Integer giving the chunk size. If left NULL, a suitable size will be chosen automatically.

cacheDir

Directory in which to save cache files. The files are deleted on normal exit but persist if the function terminates abnormally.

cacheBase

Base for the file names of cache files.

collectGarbage

Logical: should garbage collection be forced after each major calculation?

verbose

Integer level of verbosity of diagnostic messages. Zero means silent, higher values make the output progressively more and more verbose.

indent

Indentation for diagnostic messages. Zero means no indentation, each unit adds two spaces.

Details

This function calculates consensus in a hierarchical manner, using a separate (and possibly different) set of consensus options at each step. The "recipe" for the consensus calculation is supplied in the argument consensusTree.

The argument consensusTree should have the following components: (1) inputs must be either a character vector whose components match names(inputData), or consensus trees in the own right. (2) consensusOptions must be a list of class "ConsensusOptions" that specifies options for calculating the consensus. A suitable set of options can be obtained by calling newConsensusOptions. (3) Optionally, the component analysisName can be a single character string giving the name for the analysis. When intermediate results are returned, they are returned in a list whose names will be set from analysisName components, if they exist.

The actual consensus calculation at each level of the consensus tree is carried out in function consensusCalculation. The consensus options for each individual consensus calculation are independent from one another, i.e., the consensus options for different steps can be different.

Value

A list containing the output of the top level call to consensusCalculation; if keepIntermediateResults is TRUE, component inputs contains a (possibly recursive) list of the results of intermediate consensus calculations. Names of the inputs list are taken from the corresponding analysisName components if they exist, otherwise from names of the corresponding inputs components of the supplied consensusTree. See example below for an example of a relatively simple consensus tree.

Author(s)

Peter Langfelder

See Also

newConsensusOptions for obtaining a suitable list of consensus options;

consensusCalculation for the actual calculation of a consensus that underpins this function.

Examples

# We generate 3 simple matrices
set.seed(5)
data = replicate(3, matrix(rnorm(10*100), 10, 100))
names(data) = c("Set1", "Set2", "Set3");
# Put together a consensus tree. In this example the final consensus uses 
# as input set 1 and a consensus of sets 2 and 3. 

# First define the consensus of sets 2 and 3:
consTree.23 = newConsensusTree(
           inputs = c("Set2", "Set3"),
           consensusOptions = newConsensusOptions(calibration = "none",
                               consensusQuantile = 0.25),
           analysisName = "Consensus of sets 1 and 2");

# Now define the final consensus
consTree.final = newConsensusTree(
   inputs = list("Set1", consTree.23),
   consensusOptions = newConsensusOptions(calibration = "full quantile",
                               consensusQuantile = 0),
   analysisName = "Final consensus");

consensus = hierarchicalConsensusCalculation(
  individualData = data,
  consensusTree = consTree.final,
  saveConsensusData = FALSE,
  keepIntermediateResults = FALSE)

names(consensus)