Description Usage Arguments Value Author(s) See Also Examples
Creates and fits a parameterised model from experimental data, network structure and basal activity information.
1 2 3 4 5 | createModel(model_links, basal_file, data.stimulation,
data.variation = "", nb_cores = 1, inits = 1000,
perform_plots = F, precorrelate = T, method = "geneticlhs",
unused_perturbations = c(), unused_readouts = c(), MIN_CV = 0.1,
DEFAULT_CV = 0.3, model_name = "default", rearrange = "bystim")
|
model_links |
Path to the file containing the network structure, either in matrix form or in list of links form. Extension .tab expected |
basal_file |
Path to the file indicating the nodes without basal activity. Extension .dat expected. |
data.stimulation |
Path to the file containing the data in MRA_MIDAS format. Extension .csv expected. |
data.variation |
Path to the file containing the coefficient of variation for each measurement in MRA_MIDAS format. If it is not provided, the function uses the replicates of the data.stimulation file to determine a variance per probe (i.e antibody/DNA fragment/...). Extension .var expected. |
nb_cores |
Number of cores that should be used for the computation |
inits |
Number of initialisation steps which should be performed (see method for the exact meaning of this value) |
perform_plots |
Whether the distribution of the residuals and the correlation plots for the parameters deduced by correlation should be plotted or not |
precorrelate |
Whether to infer some links using a linear model instead of a random initialisation |
method |
Method to be used for the initialisation, available methods are : random : Perform a Latin Hypercube Sampling to choose inits starting points then perform a gradient descent to find the local optimum for each of those points. correlation : Deduce some parameters from the correlation between the measurements for the target node and all of its input nodes, then perform random to find the other parameters. Recommended, very efficient for small datasets. genetic : Genetic algorithm with mutation only. inits is the total number of points sampled. annealing : Simulated annealing and gradient descent on the best result. inits is the maximum number of iteration without any change before the algorithm decides it reached the best value. Use not recommended. |
unused_perturbations |
Perturbations in the dataset that should not be used |
unused_readouts |
Measured nodes in the datasets that should not be used |
MIN_CV |
Minimum coefficient of variation. |
DEFAULT_CV |
Default coefficient of variation to use when none is provided and there are no replicated in the data. |
model_name |
The name of the model is derived from the name of the data.stimulation file name. If data.stimulation is a matrix or a data.frame, 'model_name' will be used to name the model. |
rearrange |
Whether the rows should be rearranged. "no" to keep the order of the perturbations from the data file, "bystim" to group by stimulations, "byinhib" to group by inhibitions. |
An MRAmodel object describing the model and its best fit, containing the data
Mathurin Dorel dorel@horus.ens.fr
importModel, exportModel, rebuildModel
Other Model initialisation: initModel
1 2 3 4 5 6 7 | ## Not run:
model = createModel("links.tab", "basal.dat", "data_MIDAS.csv") # Produces a model for the network described in links.tab using the data in data_MIDAS.csv
model = createModel("links.tab", "basal.dat", "data_MIDAS.csv", "variation.var") # Uses the variation from a variation file
model = createModel("links.tab", "basal.dat", "data_MIDAS.csv", nb_cores = detectCores()) # Uses all cores available (with the package parallel)
model = createModel("links.tab", "basal.dat", "data_MIDAS.csv", inits = 1000000) # Uses more initialisations for a complex network
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.