mzpick: EIC peak picking

In blosloos/enviPick: Peak Picking for High Resolution Mass Spectrometry Data

EIC peak picking

Description

Peak-picking within individual EIC cluster formed by mzclust without assuming a certain peak shape. Includes a baseline subtraction step.

Usage

mzpick(MSlist, minpeak = 4, drtsmall = 20, drtfill = 10,  drttotal = 200, recurs = 4, 
weight = 2, SB = 3, SN=2, minint = 1E4, maxint = 1e+07, ended = 2, progbar = FALSE, 
from = FALSE, to = FALSE)

Arguments

MSlist	An MSlist returned by mzclust
minpeak	Peak definition - minimum number of measurements required within the RT window of drtsmall
drtsmall	Peak definition - RT window of a peak; cp. minpeak argument
drtfill	Maximum RT gap length to be filled, cp. details
drttotal	Peak definition - Maximum RT length of a single peak
recurs	Maximum number of peaks within one EIC, cp. details
weight	Weight for assigning measurements to a peak, cp. details
SB	Peak definition - signal-to-base ratio
SN	Peak definition - signal-to-noise ratio
minint	Peak definition - minimum intensity of a peak
maxint	Peaks above this intensity are always retained, regardless of other checks
ended	Within the peak detection recursion set by argument recurs, how often can a peak detection fail to end the recursion?
get_mass	Use default mean (mean) or intensity-weighted mean (wmean) of centroids to derive peak m/z?
progbar	For debugging, ignore
from	For debugging, ignore
to	For debugging, ignore

Details

In a first step, RT gaps between measurements in an EIC not larger than drtfill are filled by linear interpolation. Subsequently, peaks are assigned over a number of recurs recursions not interrupted by more than ended times of failed peak detections.

At each recursion, the most intense EIC measurement not yet assigned to a peak is selected as peak apex and neighbouring unassigned measurements at lower and higher RT are evaluated for forming the peak. To this end, increases (lower RT) and decreases (higher RT) in intensity of consecutive measurements over a maximum RT width of drtdens are summed and penalized by a factor of weight for intensity reversions. The measurements with optimum values are then selected to define the start and end measurement of the peak.

Thereupon, the candidate peak is checked to (a) have at least minpeaks within a RT window of drtsmall, (b) be larger than the minimum peak intensity minint and (c) have a minimum SB ratio (the ratio between the most intense measurement and the mimimum intensity of the first or last peak measurement). Candidate peaks failing in any of the aspects (a) to (c) are discarded (adding to ended), unless they are higher in intensity than maxint.

Next, all measurements assigned to peaks are removed from the EIC and the resulting gaps linearly interpolated and smoothed by a moving window average to form a baseline. The latter is then subtracted from the assigned peaks.

In a last step, peaks are checked for their signal-to-noise SN ratio in relation to the baseline measurements (if present). Herein, SN is defined as the ratio between the most intense (baseline-corrected) peak measurement and the median of the difference between the non-peak measurements (if any) and the baseline.

Value

Returns the argument MSlist, with entries made:

Parameters	MSlist[[2]]: saves the parameter settings.
Scans	MSlist[[4]]: matrix with raw measurements and tags for picked peaks within EICs within partitions.
Peak_index	MSlist[[7]]: Index assigning picked peaks to sections in the raw measurment of MSlist[[4]]. Required for fast access.
Peaklist	MSlist[[8]]: matrix with picked peak characteristics, with columns: m/z (mean m/z of peak measurements), var_m/z (m/z variation of peak measurements), max_int (base-line corrected maximum intensity), sum_int (sum of all base-line corrected peak measurement intensities), RT (retention time at maximum intensity), minRT (start peak RT), maxRT (end peak RT), peak# (peak ID number), EIC# (EIC ID number) and Score (not yet implemented).

imbecile

ended must be smaller than recurs. minpeak and drtsmall should be congruent in both and mzclust and mzpick.

Author(s)

Martin Loos

See Also

plotMSlist writePeaklist

blosloos/enviPick documentation built on Feb. 13, 2023, 12:03 p.m.