modelTurnover: Estimate protein/peptide turnover

In marcpaga/pulsedSilac: Analysis of pulsed-SILAC quantitative proteomics data

Description

Method to apply turnover models on protein/peptide data

Usage

modelTurnover(x, ...)

## S4 method for signature 'SilacProteinExperiment'
modelTurnover(
  x,
  assayName = "fraction",
  formula = "fraction ~ 1-exp(-k*t)",
  start = list(k = 0.02),
  robust = FALSE,
  mode = "protein",
  verbose = FALSE,
  returnModel = FALSE,
  conditionCol,
  timeCol,
  silent = TRUE,
  ...
)

## S4 method for signature 'SilacPeptideExperiment'
modelTurnover(
  x,
  assayName = "fraction",
  formula = "fraction ~ 1-exp(-k*t)",
  start = list(k = 0.02),
  robust = FALSE,
  mode = c("grouped", "peptide"),
  verbose = FALSE,
  returnModel = FALSE,
  conditionCol,
  timeCol,
  proteinCol,
  silent = TRUE,
  ...
)

## S4 method for signature 'SilacProteomicsExperiment'
modelTurnover(
  x,
  assayName = "fraction",
  formula = "fraction ~ 1-exp(-k*t)",
  start = list(k = 0.02),
  robust = FALSE,
  mode = c("protein", "grouped", "peptide"),
  verbose = FALSE,
  returnModel = FALSE,
  conditionCol,
  timeCol,
  proteinCol,
  silent = TRUE,
  ...
)

Arguments

x	A SilacProteinExperiment, SilacPeptideExperiment or SilacProteomicsExperiment object.
...	further parameters passed into nls or nlrob.
assayName	character indicating which assay to use as data input for the model.
formula	formula to be used. Time must always be named "t" and the data must be named "fraction".
start	named list with the initical values for the parameters in formula.
robust	logical indicating if robust modelling from the robustbase package should be used.
mode	character indicating which type of data should be used. Can be "protein": one model per protein; "grouped": one model per protein using peptide data; "peptide" one model per peptide.
verbose	logical indicating if a progress bar should be printed.
returnModel	logical indicating if the model objects should be returned also in the output.
conditionCol	character indicating which column of colData(x) describes the conditions.
timeCol	character indicating which column of colData(x) describes time.
silent	logical indicating if the errors given by nls/nlrob should be printed.
proteinCol	character indicating which column of rowData(x) describes the assigned protein to a peptide. (Only for peptide data)

Details

The nls and nlrob functions have many arguments that can be tunned for parameter fitting. Unfortunately, not all the possible argument combinations have been tested. It is recommended to first test one model with the desired parameters with silent = FALSE to see that it runs smoothly and then run the whole proteome with silent = TRUE to supress failed convergence errors. For example, some methods for nlrob use upper and lower bounds instead of start.

Please open an issue on github if the function is having trouble with a particular argument.

For robust modelling the method 'CM' and 'mtl' are not yet supported.

Value

A named list with either model metrics in matrices or the model objects.

Examples

data('wormsPE')
wormsPE <- calculateIsotopeFraction(wormsPE, ratioAssay = 'ratio')

modelList <- modelTurnover(x = wormsPE[1:10],
                           assayName = 'fraction',
                           formula = 'fraction ~ 1 - exp(-k*t)',
                           start = list(k = 0.02),
                           mode = 'protein',
                           robust = FALSE,
                           returnModel = TRUE,
                           silent = TRUE)

marcpaga/pulsedSilac documentation built on March 11, 2020, 8:49 p.m.