How to Use PeakFinder

Overview

PeakFinder is a universal peak detection tool for any 1D signal: chromatography, spectroscopy, qPCR melt curves, EEG, sensor data, etc. It uses a first-difference sign-change algorithm for peak detection, followed by prominence-based filtering and parabolic interpolation for sub-point precision. An optional multi-peak Gaussian fitting is available.

Algorithm Pipeline

• Step 1 - Smoothing (optional): Savitzky-Golay filter (2nd order polynomial).
• Step 2 - Baseline (optional): None / Percentile / Rolling Min.
• Step 3 - Peak Detection: First-difference sign change with platform tolerance.
• Step 4 - Prominence: Peak height minus max(left saddle, right saddle).
• Step 5 - FWHM: Full Width at Half Maximum.
• Step 6 - Parabolic Interpolation: 7-point Gaussian-weighted quadratic fit for sub-point precision.
• Step 7 - Gaussian Fitting (optional): Multi-peak Gaussian fitting using detected peaks as initial values. Max 5 peaks. Optimized via L-BFGS-B.

Output Columns

• Peak Table: Position, Height, Prominence, FWHM, Area, SE, CI, R2
• Fit Table: Initial vs fitted Position, Height, Sigma, FWHM. R2 of overall fit.

Demo Datasets

• Gaussian (Clean): 3 well-separated Gaussian peaks.
• Gaussian (Noisy): Same + noise.
• Overlapping: Two close peaks + one separate.
• Chromatogram-like: 7 peaks of varying heights and widths.
• Cell Cycle: Flow cytometry G1/S/G2+M phase distribution.
• ECG: Simulated QRS-P-T complexes.
• Seismic: P-wave, S-wave, surface wave arrivals.