Data Preview

Multi-peak (-dF/dT)

Melt Curve

Tm Values


                

All Tm Values


Multi-peak Report


            

Help & Instructions

1. Data Format

Please submit data in the same format as the demo data. Your Excel/CSV file should have:

  • First column: Temperature values (°C)
  • Subsequent columns: Fluorescence values for each well (A01, A02, B01, etc.)

2. How to Use
  • Load your data or use the built-in demo
  • Click 'Filter & Find Peaks' to run preprocessing
  • Click a well in the plate grid to see single well results
  • Wells are color-coded: Green=Passed, Orange=No Peak, Gray=No Data
  • Click 'Run Analysis (All Wells)' for batch results

3. Peak Detection Parameters

These parameters control the preprocessing step (Gaussian smoothing + Savitzky-Golay derivative + PeakDetection algorithm). Click 'Filter & Find Peaks' after changing these values.

  • Min signal range (RFU): Minimum acceptable fluorescence range (max − min). Wells with a smaller swing are tagged as No Peak (retained in the plate grid and batch output, but no Tm reported). Default 1500. Lower values accept weaker signals; higher values require stronger amplitude.
  • Peak detection window (points): The moving-window size used to scan the −dF/dT curve for potential peak regions. A larger window makes the detector less sensitive to narrow noise spikes but may miss genuinely sharp peaks. Default 7.
  • SG smoothing window (points): Savitzky-Golay smoothing window used for derivative calculation. Auto-set to detection window + 1 (e.g., 8 when detection = 7). Larger values give smoother derivatives but can blur closely spaced peaks. Read-only.
  • Consecutive windows (N): How many consecutive detection windows must exceed the threshold to declare 'peak present'. Larger values make the detector more conservative (fewer false positives) but may miss weak or narrow peaks. Default 2.
  • Significance multiplier (k): The PeakDetection threshold = k × median(window_std). This is the key sensitivity knob. A higher k makes the detector stricter (fewer wells pass, fewer shoulder peaks); a lower k is more permissive. Default 1.4. Typical range 1.0–3.0.

4. Peak Refinement Parameters

These filters are applied to the candidate peaks after PeakDetection has declared a well 'has peak'. They do not affect preprocessing — you can change them and click a well or 'Run Analysis' without re-running 'Filter & Find Peaks'.

  • Min peak height (% of max): Minimum height of a candidate peak relative to the tallest peak in that well. If the tallest −dF/dT value is 1000, a 5% threshold means peaks below 50 are dropped. Default 5%.
  • Min sharpness (% of max): Drop flat shoulders by comparing how 'pointy' each peak is. For every candidate peak, a small window (±7 data points, about ±2°C) around the peak is taken. Inside this window, the mean squared slope of the −dF/dT curve is computed. This value is then divided by the sharpest peak's score to obtain a percentage. Peaks below this percentage are discarded. Default 10%. Typical values: shoulders score 2–5%, real peaks score 30–100%.
  • Min Tm distance (°C): When two peaks are closer than this, they are merged and only the taller one is kept. Use this to suppress split peaks or fine structure that should be treated as a single Tm. Default 1.5°C.
  • Min peak width (°C): Minimum half-width at base (distance from left valley to right valley, divided by 2). Very narrow spikes (often noise) are dropped. Default 0.5°C.

5. Output
  • Single Well: Multi-peak (−dF/dT) plot + Melt curve (original + Tm lines) + Tm text (with CI and SE)
  • Batch Results: Table of Tm values (up to 6 peaks per well) with confidence intervals and standard errors; multi-peak text report
  • Plate Grid: Deep green = Passed (has peak), Orange = No Peak (low signal or no peak detected), Gray = No Data
  • Download: Excel (.xlsx) with Tm + CI + SE for all wells; PNG of all melt curves

6. Workflow Summary

Step 1: Load data (or use demo). Step 2: Adjust detection parameters if needed. Step 3: Click 'Filter & Find Peaks' to run preprocessing and see plate grid colors. Step 4: Click individual wells to inspect single-well plots. Step 5: Adjust refinement parameters (height/sharpness/distance/width) if shoulders or noise peaks remain. Step 6: Click 'Run Analysis (All Wells)' for batch results and download.