Gaussian Peak Shape Evaluation

Gaussian Peak Shape Evaluation

Summary

Quantifies the goodness-of-fit of detected LC-MS peaks to a Gaussian model to filter out poorly shaped peaks and retain high-quality features. This metric serves as one of three orthogonal quality thresholds (alongside SNR and prominence) in asari's peak quality filtering pipeline.

When to use

After peak detection on a composite mass track has identified candidate peaks in a mass chromatogram, and before compiling the final feature table. Apply this skill when you need to remove peaks with asymmetric, bimodal, or otherwise non-Gaussian elution profiles that may indicate co-elution, baseline noise, or instrumental artifacts rather than true analyte signals.

When NOT to use

• Input peaks are already known to be high-quality standards or reference compounds (skip Gaussian evaluation and use direct intensity matching).
• Data are from GC-MS workflows where peak shape expectations differ from LC-MS Gaussian assumptions; alternative shape models may be required.
• Peak detection has not yet been performed; this skill applies only after candidate peaks have been identified, not during initial m/z binning or mass track construction.

Inputs

• Composite map peak detection output (JSON or structured format) containing candidate peaks with m/z, retention time, intensity array, peak height, SNR, and prominence
• Detected peak intensity list (array of intensity values across scan range for single peak)

Outputs

• Goodness-of-fit score (numeric, range 0–1 or similar correlation metric) per peak
• Binary pass/fail classification for each peak (goodness_fitting > 0.5)
• Filtered peak list subset passed to feature table compilation

How to apply

Extract intensity values from the detected peak region (typically 5–10 scan points around the apex) for each candidate peak. Fit a Gaussian curve (via least-squares optimization or similar) to the intensity array and compute the goodness-of-fit metric (R² or similar correlation coefficient). Retain only peaks where goodness_fitting > 0.5 (default threshold). This threshold balances rejection of severely deformed peaks against over-filtering of realistic chromatographic variation. Combine this filter with parallel SNR (>2) and peak height/prominence thresholds; a peak must pass all three to enter the final feature table. The rationale is that Gaussian peaks indicate clean analyte elution, while non-Gaussian profiles suggest contamination or interference.

Related tools

• asari peaks module (Implements evaluate_gaussian_peak_on_intensity_list() function to fit Gaussian curve and compute goodness-of-fit metric on intensity arrays) — https://github.com/shuzhao-li/asari
• scipy.signal.find_peaks (Detects local maxima and prominence on mass tracks prior to Gaussian shape evaluation; provides candidate peaks and prominence values passed to shape filter)

Examples

from asari.peaks import evaluate_gaussian_peak_on_intensity_list; goodness = evaluate_gaussian_peak_on_intensity_list(intensity_array); peak_passes = goodness > 0.5

Evaluation signals

• Goodness-of-fit scores for all peaks are in the valid range (0–1 or correlation coefficient bounds); no NaN or inf values.
• Peak count after shape filter is consistent with expected reduction (typically 20–50% of unfiltered peaks depending on sample complexity and instrument calibration).
• Peaks below goodness_fitting threshold show visibly asymmetric or multi-modal intensity profiles when plotted; peaks above threshold are unimodal Gaussian-like.
• Row count reduction from full_Feature_table.tsv (unfiltered) to preferred_Feature_table.tsv (filtered with SNR, shape, and height thresholds) is proportional and documented in project output.
• Manually inspected mass tracks (via dashboard or export) for a sample of rejected and retained peaks confirm that retained peaks have cleaner, more symmetric elution profiles.

Limitations

• Goodness-of-fit threshold (0.5) is a fixed default and may not generalize across different LC instruments, columns, or flow rates where peak broadness and shape vary naturally.
• Peaks with shouldering or multiplet structure due to true co-elution (e.g., isomers) may fail the Gaussian filter even though they represent real analytes; no automatic distinction between instrumental artifact and biological co-elution is made.
• Very narrow peaks (single or few scans) may have high apparent goodness-of-fit by chance; this skill does not enforce minimum peak width, relying instead on the minimum peak height and prominence filters to exclude noise spikes.
• The Gaussian model assumes symmetric, unimodal peaks and does not account for tailing or fronting common in reversed-phase LC; alternative shape models may be needed for specialized chromatography modes.

Evidence

• [other] Apply peakshape threshold filter (goodness_fitting > 0.5) using gaussian peak evaluation to retain well-shaped peaks.: "Apply peakshape threshold filter (goodness_fitting > 0.5) using gaussian peak evaluation to retain well-shaped peaks."
• [methods] See peaks.evaluate_gaussian_peak_on_intensity_list: "See peaks.evaluate_gaussian_peak_on_intensity_list"
• [other] Asari applies peak quality filtering by tracking selectivity metrics on m/z, chromatography, and annotation databases to refine detected features after composite map peak detection.: "Asari applies peak quality filtering by tracking selectivity metrics on m/z, chromatography, and annotation databases to refine detected features after composite map peak detection."
• [readme] All peaks are kept in export/full_Feature_table.tsv if they meet signal (snr) and shape standards: "All peaks are kept in export/full_Feature_table.tsv if they meet signal (snr) and shape standards"
• [intro] Statistics guided peak dection, based on local maxima and prominence, selective use of smoothing: "Statistics guided peak dection, based on local maxima and prominence, selective use of smoothing"