generate_synthetic_data: Generate a data set according to the probabilistic dropout...

In const-ae/proDD: Identifying Differentially Abundant Proteins from Label-Free Mass Spectrometry Data

Description

Specify the number of rows in the dataset, the number of conditions and replicates, how many proteins have a different mean and a few additional hyperparameters and get a synthetic dataset the is similar to data from a real label-free mass spectrometry experiment.

Usage

generate_synthetic_data(n_rows, experimental_design = NULL,
  n_replicates = as.numeric(table(experimental_design)),
  n_conditions = length(n_replicates), frac_changed = 0.1,
  n_changed = round(n_rows * min(1, frac_changed)), mu0 = 20,
  sigma20 = 10, nu = 3, eta = 0.3, rho = rep(18, times = if
  (length(n_replicates) == 1) n_replicates * n_conditions else
  sum(n_replicates)), zeta = rep(-1, times = if (length(n_replicates) ==
  1) n_replicates * n_conditions else sum(n_replicates)))

Arguments

n_rows	integer. The number of rows in the new dataset
experimental_design	a vector that specifies which samples belong to the same condition. Default: 'NULL' in which case 'n_replicates' must be specified
n_replicates	integer or vector. The number of replicates in each condition.
n_conditions	The number of conditions. Setting 'n_replicates=3' and 'n_conditions=2' is equal to specifying 'experimental_design=c(1,1,1,2,2,2)'.
frac_changed	the fraction of rows for which different means are drawn for each conditon.
n_changed	alternative way to specify for how many rows have different means in each condition.
mu0	the global mean around which the row means are drawn. Default '20'
sigma20	the global variance specifying the spread of means around 'mu0'. Default '10'.
nu	degrees of freedom for the the global variance prior. Default '3'.
eta	scale of the global variance prior. Default '0.3'.
rho	vector specifying the intensity where the chance of a dropout is 50/50. Either length one or same length as 'n_replicates * n_conditons' or 'length(experimental_design)' respectively. Default '18'.
zeta	vector specifying the scale of the dropout curve. Either length one or same length as 'n_replicates * n_conditons' or 'length(experimental_design)' respectively. Default '18'.

Value

a list with 5 elements

X

the data matrix with missing values

t_X

the true data matrix, before data dropped out

mus

matrix of size 'n_rows * n_conditions'. The true means for each condition

sigmas2

a vector of size 'n_rows'. The true variance for each row.

changed

a boolean vector of size 'n_rows', with the label if a row has different means for each condition

Examples

 data <- generate_synthetic_data(n_rows=10,
                n_replicates=3, n_conditions=2)

 data2 <- generate_synthetic_data(n_rows=10,
                experimental_design=c(1,1,1,2,2,2))

 data3 <- generate_synthetic_data(n_rows=10,
                rep(letters[1:3], each=4))

const-ae/proDD documentation built on Jan. 14, 2020, 9:34 a.m.