netbenchmark: Benchmarking of several network inference algorithms from...

In netbenchmark: Benchmarking of several gene network inference methods

Description

For a given vector of character of the names of wrapper functions that compute a network inference methods, netbenchmark performs a benchmark between them. It makes use of four different big gene datasources, it relies on a random subsampling without repetition of each one of the datasets and noise addition in order to generate the source data.

Usage

    netbenchmark(methods="all.fast",datasources.names="all",experiments=150,
        eval="AUPR",no.topedges=20,datasets.num=5,local.noise=20,
        global.noise=0,noiseType="normal",sym=TRUE,plot=FALSE,seed=NULL,
        verbose=TRUE)

Arguments

methods	A vector of characters containing the names of network inference algorithms wrappers to be compared (default: "all.fast").
datasources.names	A vector of characters containing the names of network datasources to be included in the benchmark (default: "all").
experiments	Integer specifying the number of experiments to generate the subsampled datasets (default: 150) - see datasource.subsample.
eval	The name of the evaluation metric among the following ones: "no.truepos", "AUROC" or "AUPR" (default : "AUPR") - see evaluate.
no.topedges	Float specifying the percentage number of links to be considered in the evaluation (default: 20).
datasets.num	Integer specifying the number of datasets.num to be generated for each of the selected original datasources (default: 5).
local.noise	Integer specifying the desired percentage of local noise to be added at each of the subsampled datasets (default: 20) - see datasource.subsample.
global.noise	Integer specifying the desired percentage of global noise to be added at each of the subsampled datasets (default: 20) - see datasource.subsample.
noiseType	Character specifying the type of the noise to be added: "normal" or "lognormal" (default: "normal") - see datasource.subsample.
sym	Logical specifying if the evaluation is symmetric (default: TRUE) - see evaluate.
plot	(default: FALSE)
return.nets	(default: FALSE)
seed	A single value, interpreted as an integer to specify seeds, useful for creating simulations that can be reproduced (default: NULL) - see set.seed.
verbose	Logical specifying if the code should provide a log about what the function is doing (default: TRUE).

Details

The argument methods accepts "all.fast" and "all" (case insensitive) as a parameters:

• "all.fast" performs network inference with "aracne", "c3net", "clr", "GeneNet", "mutual ranking", "mrnet", "pcit" (and registered methods with RegisterWrapper.)

• "all" performs network inference with "aracne", "c3net", "clr", "GeneNet", "Genie3", "mutual ranking", "mrnet", "mrnetb", "pcit" (and registered methods with RegisterWrapper.)

The argument datasources.names accepts "all" or a selection of the following datasources Availabledata:

• "rogers1000"

• "syntren300"

• "syntren1000"

• "gnw1565"

• "gnw2000"

All the measures only evaluates the first no.topedges % of the possible links inferred by each algorithm at each dataset. The statistical used is the Wilcoxon Rank Sum Test (wilcox.test). This test compares the number of true positives of any method with number of trials specified with the best method at each replicate.

Value

netbenchmark returns a list with six elements.

1. A data.frame which is the result table of the selected measure.

2. A data.frame which is the corresponding pvalue table of the corresponding statistical test for each one of the datasets.num between the best algorithm and the others.

3. A data.frame that sumarizes the first data.frame presenting the mean and standard deviation of the measures of each algorithm per datasource.

4. A data.frame which contains the CPU Time Used (in seconds) by the algorithm to infer the network.

5. A list containing the mean precision recall curves of the different algorithms for each datasource.

6. The seed of the random number generators that allows the replication of the results.

Each of these data.frame will have the same number of columns as methods provided by the user and an additional one for a random method, and the number of rows will depend on the number of datasets.num and datasources.name specified by the user.

Author(s)

Pau Bellot, Catharina Olsen and Patrick E Meyer Maintainer: Pau Bellot <[e-mail]>

See Also

datasource.subsample, evaluate, comp.metr

Examples

    top20.aupr <- netbenchmark(methods="all",datasources.names = "Toy",
                               local.noise=20,global.noise=10,
                               noiseType=c("normal","lognormal"),
                               datasets.num = 2,experiments = 40)
    ## Not run: 
        # Other possible studies 
        top20.fast.list <- netbenchmark()
        top20.list <- netbenchmark(methods="all",eval="no.truepos")
        top50.auroc.list <- netbenchmark(datasets.num=8,eval="AUROC",
            no.topedges=50,global.noise=10)
        top9.list <- netbenchmark(datasets.num=8,no.topedges=9,local.noise=15,
            noiseType="lognormal")
        #To export the tables to LaTeX 
        # library(xtable)
        # xtable(top20.fast.list[[1]])
    
## End(Not run)

netbenchmark documentation built on April 28, 2020, 7 p.m.