fit_parameters: Fit the probabilistic dropout parameters

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

Description

The method infers the position and scale of the dropout sigmoids, the location prior of the means and the prior for the variance. In addition it estimates some feature parameters (mean, uncertainty of mean and variance for each protein and condition).

Usage

fit_parameters(X, experimental_design, dropout_curve_calc = c("sample",
  "global_scale", "global"), frac_subsample = 1,
  n_subsample = round(nrow(X) * frac_subsample), max_iter = 10,
  epsilon = 0.001, verbose = FALSE)

## S4 method for signature 'SummarizedExperiment'
fit_parameters(X, experimental_design,
  dropout_curve_calc = c("sample", "global_scale", "global"),
  frac_subsample = 1, n_subsample = round(nrow(X) * frac_subsample),
  max_iter = 10, epsilon = 0.001, verbose = FALSE)

## S4 method for signature 'MSnSet'
fit_parameters(X, experimental_design,
  dropout_curve_calc = c("sample", "global_scale", "global"),
  frac_subsample = 1, n_subsample = round(nrow(X) * frac_subsample),
  max_iter = 10, epsilon = 0.001, verbose = FALSE)

Arguments

X	the numerical data where each column is one sample and each row is one protein. Missing values are coded as NA.
experimental_design	a vector that assignes each sample to one condition. It has the same length as the number of columns in X. It can either be a factor, a character or a numeric vector. Each unique element is one condition. If X is a SummarizedExperiment or an MSnSet object, experimental_design can also be the name of a column in the colData(X) or pData(X), respectively.
dropout_curve_calc	string that specifies how the dropout curves are estimated. There are three different modes. "sample": number of curves= number of samples, "global": number of curves=1, "global_scale": estimate only a single scale of the sigmoid, but estimate the position per sample. Default: "sample".
frac_subsample	number between 0 and 1. Often it is not necessary to consider each protein, but the computation can be significantly sped up by only considering a subset of the subsets. Default: 1.0.
n_subsample	number between 1 and nrow(X). Alternative way to specify how many proteins are considered for estimating the hyper-parameters. Default: nrow(X) * frac_subsample.
max_iter	integer larger than 1. How many iterations are run at most trying to reach convergence. Default: 10.
epsilon	number larger than 0. How big is the maximum remaining error for the algorithm to be considered converged. Default: 10^-3
verbose	boolean. Specify how much extra information is printed while the algorithm is running. Default: FALSE.

Value

a list containing the infered parameters. The list is tagged with the class "prodd_parameters" for simpler handling in downstream methods

Methods (by class)

• SummarizedExperiment: S4 method of fit_parameters for SummarizedExperiment

• MSnSet: S4 method of fit_parameters for MSnSet

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