selectAnchors: Select Anchors for Nonlinear RT Model

View source: R/selectAnchors.R

selectAnchorsR Documentation

Select Anchors for Nonlinear RT Model

Description

A subset of possible alignments in the combinedTable are used as ordered pairs to anchor a retention time projection model. Alignments of abundant features are prominent targets for anchor selection, but shared identified features (i.e. feature pairs where idx = idy) may be used.

Usage

selectAnchors(
  object,
  useID = FALSE,
  tolmz = 0.003,
  tolQ = 0.3,
  tolrtq = 0.3,
  windx = 0.03,
  windy = 0.03,
  brackets_ignore = c("(", "[", "{")
)

Arguments

object

metabCombiner object.

useID

logical. Option to first search for IDs as anchors.

tolmz

numeric. m/z tolerance for prospective anchors

tolQ

numeric. Quantile Q tolerance for prospective anchors

tolrtq

numeric. Linear RT quantile tolerance for prosepctive anchors.

windx

numeric. Retention time exclusion window around each anchor in X dataset. Optimal values are between 0.01 and 0.05 min (1-3s)

windy

numeric. Retention time exclusion window around each anchor in dataset Y. Optimal values are between 0.01 and 0.05 min (1-3s)

brackets_ignore

If useID = TRUE, bracketed identity strings of the types included in this argument will be ignored.

Details

In order to map between two sets of retention times, a set of ordered pairs need to be selected for the spline fit. This function relies on mutually abundant features to select these ordered pairs. In iterative steps, the most abundant (as indicated by Q value) in one dataset is selected along with its counterpart, and all features within some retention time window specified by windx & windy arguments are excluded. This process is repeated until all features have been considered.

tolQ & tolmz arguments restrict to feature pairs that have differences in Q & m/z within these tolerances. tolrtq further limits to feature pairs those with relative differences in linear retention time quantiles, calculated as rtqx = (rtx - min(rtx)) / (max(rtx) - min(rtx)) & rtqy = (rty - min(rty)) / (max(rty) - min(rty))

Shared identities (in which idx & idy columns have matching, non-empty & non-bracketed strings) may be used if useID is set to TRUE. In this case, shared identities will be searched first and will not be subject to any of the restrictions in m/z, Q, or rt. The iterative process proceeds after processing of shared identities.

Value

metabCombiner object with updated anchors slot. This is a data.frame of feature pairs that shall be used to map between retention times using a GAM or LOESS model.

idx

identities of features from dataset X

idy

identities of features from dataset Y

mzx

m/z values of features from dataset X

mzy

m/z values of features from dataset Y

rtx

retention time values of features from dataset X

rty

retention time values of features from dataset Y

rtProj

model-projected retention time values from X to Y

Qx

abundance quantile values of features from dataset X

Qy

abundance quantile values of features from dataset Y

adductX

adduct label of features from dataset X

adductY

adduct label of features from dataset Y

group

m/z feature group of feature pairing

labels

anchor labels; "I" for identity, "A" for normal anchors

See Also

getAnchors, fit_gam, fit_loess

Examples

data(plasma30)
data(plasma20)

p30 <- metabData(plasma30, samples = "CHEAR")
p20 <- metabData(plasma20, samples = "Red", rtmax = 17.25)
p.comb <- metabCombiner(xdata = p30, ydata = p20, binGap = 0.005)

##example 1 (no known IDs used)
p.comb <- selectAnchors(p.comb, tolmz = 0.003, tolQ = 0.3, windx = 0.03,
    windy = 0.02, tolrtq = 0.3)

##example 2 (known IDs used)
p.comb <- selectAnchors(p.comb, useID = TRUE, tolmz = 0.003, tolQ = 0.3)

##To View Plot of Ordered Pairs
anchors = getAnchors(p.comb)
plot(anchors$rtx, anchors$rty, main = "Selected Anchor Ordered Pairs",
    xlab = "rtx", ylab = "rty")


hhabra/Combiner documentation built on Jan. 26, 2024, 10:30 p.m.