detect_rel_bias: Detect systematic bias in metabolomic time-course data

In ssokolen/metcourse: Tools for Metabolic Time-course Data

Detect systematic bias in metabolomic time-course data

Usage

detect_rel_bias(time, concentration, metabolite, min.deviation = NULL, ...)

Arguments

time	A vector of times or sample numbers for metabolite time-courses. Note, there should be a time value for every concentration for every metabolite e.g. time = c(1, 2, 3, 1, 2, 3), concentration = c(20, 15, 10, 3, 6, 9), metabolite = c('glc', 'glc', 'glc', 'lac', 'lac', 'lac').
concentration	A vector of metabolite concentrations.
metabolite	A vector of metabolite names that correspond to time and concentration.
min.deviation	Smallest median relative deviation to identify as a bias. If not supplied, it will be estimated as 50 \item ...Arguments to be passed into fit_b_spline, (such as degree and number of knots).

A dataframe corresponding to the time points with and without systematic error. Identifies systematic deviations in metabolic data using a B-spline fit. Timepoints that have a median relative deviation above a threshold value are assumed to be influenced by a measurement or methodological bias as compared to the overall trends metabolite concentrations.

# Using previously simulated data 40 metabolic trends with 10 time points data(timecourse)

# Adding an error of 5% at sample 4 logic <- timecourse$sample == 4 timecourse$concentration[logic] <- timecourse$concentration[logic] * 1.05

# Estimating error <- correct_rel_bias(timecourse$time, timecourse$concentration, timecourse$metabolite)

ssokolen/metcourse documentation built on Nov. 3, 2024, 11:22 p.m.