Metabolomic Peak Quality Assessment

metabolomic-peak-quality-assessment

Summary

Evaluate detected peaks from LC/HRMS data using quantitative chromatographic properties and signal quality metrics to filter peaks meeting sensitivity, specificity, and reproducibility standards. This skill ensures high-confidence peak lists suitable for population-scale untargeted metabolomics studies.

When to use

After peak detection in untargeted LC/HRMS workflows, when you have a list of candidate peaks with signal intensity profiles and need to filter them according to data quality thresholds (signal-to-noise ratio, peak width, baseline separation, and isotopic pairing consistency) before downstream alignment and annotation.

When NOT to use

• Input peaks already pass through a different peak quality filter (e.g., from MZmine 2 or xcms vendor workflow) — applying IDSL.IPA's evaluation may introduce redundant or conflicting quality criteria.
• Data are targeted analysis with pre-defined m/z-RT pairs — use IDSL.IPA's IPA_targeted function instead of the full EIC-to-peak-detection-to-evaluation pipeline.
• Chromatographic method or instrument differs substantially from supported formats (mzXML, mzML, netCDF) or produces atypical peak shapes for which published metric thresholds do not apply.

Inputs

• Detected peak list (retention time, m/z, apex intensity, detection confidence scores)
• EIC signal intensity profiles (continuity and intensity transitions per candidate)
• Mass spectrometry acquisition parameters (scan rate, baseline noise estimate)

Outputs

• Filtered peak list with retention time, m/z, peak area, and 19 computed chromatographic properties
• Quality assessment report with per-peak metrics (S/N, peak width, asymmetry, gaussianity, nIsoPair)
• Gap-filled peak height tables and pairwise correlation lists for adduct/fragment detection

How to apply

Apply IDSL.IPA's peak property evaluation algorithm to compute 19 chromatographic metrics per detected peak, including signal-to-noise ratio (using baseline, xcms, or RMS methods), peak width, asymmetry factor, USP tailing factor, gaussianity, and nIsoPair/RCS for isotopic coherence. Define cutoff thresholds for each metric based on your chromatographic method and mass spectrometry platform (e.g., S/N > baseline threshold, peak width within expected range, tailing factor near 1.0 for symmetric peaks). Filter retained peaks by applying these thresholds sequentially; peaks failing any criterion are excluded. Rationale: multi-dimensional quality filtering reduces false positives and improves alignment reliability across population-size studies by selecting only peaks with good peak shape, sufficient signal above noise, and consistent isotopic signatures.

Related tools

• IDSL.IPA (Implements peak property evaluation, calculates 19 chromatographic metrics, applies quality filtering thresholds, and outputs filtered peak lists with confidence scores) — https://github.com/idslme/IDSL.IPA
• R (Execution environment for IDSL.IPA package and statistical analysis of peak quality metrics)
• IDSL.UFA (Downstream molecular formula annotation tool integrated with IDSL.IPA filtered peak lists) — https://github.com/idslme/IDSL.UFA
• IDSL.CSA (Composite spectra generation by clustering recurring ions from quality-filtered peaks) — https://github.com/idslme/IDSL.CSA

Examples

library(IDSL.IPA)
IPA_workflow("path/to/IPA_parameters.xlsx")

Evaluation signals

• Filtered peak list row count < input candidate count; verify that exclusion rate is consistent with thresholds applied (e.g., typical metabolomics studies report 50–80% retention after quality filtering).
• All retained peaks have S/N ≥ specified baseline threshold and peak width within biologically plausible range (e.g., 0.01–2 min for typical LC methods).
• Asymmetry factor and USP tailing factor values cluster near 1.0 for retained peaks; heavily tailed or fronted peaks are excluded or flagged.
• nIsoPair and RCS metrics confirm isotopic coherence (e.g., ¹³C isotologues detected with expected mass offset and intensity ratio ≈ 1.1× per carbon).
• Peaks retained after filtering show improved alignment consistency and reduced false positives in downstream cross-sample peak alignment compared to unfiltered candidate list.

Limitations

• Peak quality metric thresholds are tunable but require instrument-specific calibration; default parameters in IPA_parameters.xlsx may not generalize to all LC/HRMS platforms or chromatographic methods.
• Highly overlapping or co-eluting peaks may be incorrectly flagged as poor quality even if biologically relevant; visual inspection or targeted EIC review recommended for borderline cases.
• S/N calculation methods (baseline, xcms, RMS) yield different results; consistent method choice must be specified in PARAM configuration and documented.
• The 19-metric evaluation assumes standard metabolite mass range and peak shape; atypical compounds (very lipophilic, highly charged, or fragile metabolites) may not meet published thresholds.
• No built-in changelog documented; version-to-version threshold or metric changes may not be transparent.

Evidence

• [readme] Peak property evaluation algorithm with 19 metrics: "Calculating 19 chromatographic peak properties such as peak area, nIsoPair, RCS, cumulated intensity, R13C, peak width, RPW, number of separation trays, asymmetry factor, USP tailing factor, skewness"
• [intro] Position and function in workflow: "algorithms covering extracted ion chromatogram (EIC) candidate generation, peak detection, peak property evaluation"
• [other] Quality filtering based on signal-to-noise and peak width: "Filter detected peaks according to algorithm-defined quality criteria (signal-to-noise ratio, peak width constraints, baseline separation)."
• [other] Output includes confidence scores: "Output detected peak list with retention time, m/z, peak apex intensity, and detection confidence scores."
• [readme] Population-scale application and performance: "IDSL.IPA generates comprehensive and high-quality datasets from untargeted analysis of organic small molecules for population-size studies."
• [readme] S/N definition variants: "Definition of Signal to Noise ratio (S/N)"