Ftms Mass Spectrum Peak Detection

FT-MS Mass Spectrum Peak Detection

Summary

Automated identification and localization of mass spectral peaks in Fourier-transform ion cyclotron resonance (FT-ICR) mass spectrometry data using apex quadratic fitting. This skill is essential for converting raw FT-MS transients into interpretable peak lists suitable for downstream calibration and molecular formula assignment.

When to use

Apply this skill when you have loaded an FT-ICR raw spectrum (e.g., ESI_NEG_SRFA.d in Bruker or ThermoFisher .raw format) and need to identify the m/z positions and intensities of individual mass spectral peaks. The skill is triggered after noise threshold calculation (manual or automatic) and precedes mass calibration and molecular formula search workflows.

When NOT to use

• Input is already a centroided peak list or pre-processed mass list; peak detection has already been performed.
• The spectrum is severely baseline-corrupted or contains unresolved multiplets where quadratic fitting cannot reliably separate overlapping peaks.
• Profile-mode data from very low resolving power instruments (R < 10,000) where individual peaks are not resolved from noise.

Inputs

• FT-ICR raw spectrum (Bruker .d transient, .ser, or .fid; ThermoFisher .raw; or other vendor format supported by CoreMS)
• Noise threshold value (scalar, automatically calculated or user-defined)
• Mass spectrum object (MassSpectrum instance from CoreMS encapsulation)

Outputs

• Peak list (list of MassSpectrumPeak objects with m/z, intensity, resolving power, and quadratic fit parameters)
• Mass spectral peak table (pandas DataFrame or serialized output with peak metadata)

How to apply

After loading the mass spectrum using CoreMS data encapsulation modules and applying noise threshold filtering, invoke the peak-picking algorithm with apex quadratic fitting. This method iterates through the centroided or profile-mode spectrum, identifies local maxima that exceed the noise threshold, and fits a quadratic polynomial to the apex region to refine the m/z position and intensity estimate. The quadratic fitting improves localization accuracy compared to simple centroid or maximum-intensity methods, particularly important for high-resolving-power FT-ICR data where peak shape carries mass information. The algorithm outputs a list of detected peaks with their m/z values, intensities, and estimated resolving power.

Related tools

• CoreMS (Provides data encapsulation, noise thresholding, and peak-picking API (apex quadratic fitting implementation)) — https://github.com/EMSL-Computing/CoreMS
• numpy (Numerical computation and polynomial fitting (quadratic coefficients))
• pandas (Peak list serialization and tabular peak metadata export)

Examples

from corems.encapsulation.factory.parameters import MSParameters; from corems.data_source.raw_data import RawDataReader; spectrum = RawDataReader('tests/tests_data/ftms/ESI_NEG_SRFA.d').get_mass_spectrum(0); spectrum.peaks = spectrum.pick_peaks()

Evaluation signals

• Peak count is within expected range for spectrum complexity (e.g., >50 peaks for complex natural organic matter, <20 for simple synthetic mixtures).
• Detected peak m/z values fall within the instrument's operational mass range and do not contain spurious peaks at extreme m/z or intensity values.
• Quadratic fit residuals are small (typically <1 ppm error) compared to subsequent mass calibration errors.
• Resolving power estimates from peak shape are consistent with instrument specifications (e.g., >100,000 FWHM for FT-ICR at m/z 400 in 7 Tesla field).
• Peak intensity distribution is reasonable (e.g., maximum peak intensity >>noise threshold, intensity distribution follows expected mass spectral patterns).

Limitations

• Quadratic fitting assumes approximately symmetric peak shapes; asymmetric or tailed peaks typical of certain ion sources may have slightly degraded m/z accuracy.
• Severely overlapping or unresolved multiplets are not resolved by this algorithm; additional deconvolution workflows may be required.
• Baseline artifacts, electronic noise spikes, or contamination peaks can be mistaken for real m/z signals if noise thresholding is set too low.
• The method requires careful tuning of noise threshold; too-high thresholds miss weak peaks, too-low thresholds introduce noise peaks.

Evidence

• [readme] Peak picking using apex quadratic fitting: "Peak picking using apex quadratic fitting"
• [other] Load the mass spectrum from ESI_NEG_SRFA.d and the reference m/z file SRFA.ref using CoreMS data encapsulation modules.: "Load the mass spectrum from ESI_NEG_SRFA.d and the reference m/z file SRFA.ref using CoreMS data encapsulation modules"
• [readme] Manual and automatic noise threshold calculation: "Manual and automatic noise threshold calculation"
• [readme] Apodization, Zerofilling, and Magnitude mode FT: "Apodization, Zerofilling, and Magnitude mode FT"
• [readme] CoreMS is a comprehensive mass spectrometry framework for software development and data analysis of small molecules analysis: "CoreMS is a comprehensive mass spectrometry framework for software development and data analysis of small molecules analysis"