gLVgenerateData: Generate random data for gLV fitting

In lkshrsch/gLVInterNetworks: Inference of interaction networks based on generalised Lotka Volterra dynamics

Description

Generates random data simulating time series of cell abundances governed by Lotka Volterra dynamics

Usage

gLVgenerateData(species, number_of_interactions, timepoints, noise, testData)

Arguments

species	Integer describing the number of independent cellular subcommunities
number_of_interactions	Integer describing the number of non-zero interactions present in the simulated system. These are assigned randomly between the nodes or subcommunities
timepoints	Numeric vector containing the timepoints for which to compute the solutions of the model
noise	The standard deviation of the normally distributed stochastic factor added to the solution of the model at each time step
testData	Number of observations used as test dataset for validation on untrained data, taken from the last measurements of the time series.

Value

Returns a matrix. The first column displays the time points, and the remaining columns correspond each to a independent variable in the system.

Author(s)

Lukas Hirsch

Examples

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (species, number_of_interactions, timepoints, 
    noise) 
{
    "requires inSilico_bio, discrete, addNoise_res"
    raw_data <- inSilico_bio(species, number_of_interactions)
    Parms <- raw_data[[2]]
    res <- raw_data[[1]]
    threshold = 3
    if (any(which(round(diff(abs(rowMeans(res[, -1])), 2), threshold) == 
        0))) {
        end <- min(which(round(diff(abs(rowMeans(res[, -1])), 
            2), threshold) == 0))
    }
    else {
        end <- nrow(res)
    }
    res <- solveLV_bio(Parms, seq(0, end, by = 0.01), res[1, 
        -1])
    discretization <- nrow(res)/timepoints
    res <- discrete(res, discretization)
    test_data <- res[(nrow(res) - 19):nrow(res), ]
    obs <- res[1:(nrow(res) - 20), ]
    obs <- addNoise_res(obs, noise)
    timepoints <- nrow(obs)
    dimensions <- ncol(obs[, -1])
    k <- sum(any(Parms == 0))/length(Parms)
    data <- list(species = species, timepoints = timepoints, 
        Parms = Parms, noise = noise, sparsity = k, obs = obs, 
        testData = test_data)
    class(data) <- "Sim_data"
    return(data)
  }

lkshrsch/gLVInterNetworks documentation built on May 21, 2019, 7:33 a.m.