TFE uses a unique approach to rapidly reintegrate samples and annotations on demand. TFE caches XICs in memory through a process called extraction. These cached XICs can then be reintegrated whenever integration settings are changed. TFE also uses status colors to indicate whether an annotation or sample needs to be reextracted, reintegrated, or has incompatible settings.

Extraction

Extraction is the process of opening a raw file, pulling XIC information, and caching it in memory. This step usually takes 0.1–0.5 seconds per sample, depending on raw file size, file format, and disk performance. If peak RTs are consistent from study to study and the mass axis is calibrated, extraction only needs to be done once per study.

The extraction settings are:

  • MZ - mz - Mass-to-charge ratio of the XIC. For QQQ data, this is the precursor m/z.
  • RT - rt - Center of the XIC and expected RT of the feature.
  • PPM - ppm_window - 10 - The ± m/z extraction window for the XIC. Not used for QQQ data.
  • Range - extraction_range - 1.2 - The ± RT extraction range for the XIC.
  • Product - product - 0 - For QQQ data, this is the product m/z. Leave as 0 for ToF and Orbitrap MS1 data.
  • Polarity - polarity - -1 - 0: negative, 1: positive, -1: unspecified.

In the Popup Chart, the displayed XIC is automatically updated when extraction settings are changed. However, if you click Apply to All, TFE may or may not reextract all samples depending on the Automatically Reextract setting on the Settings page.

Integration

Integration settings operate on the extracted XICs to find integration bounds. The settings, corresponding annotations.csv headers, and recommended HPLC values for a 12-minute gradient are:

  • Smoothing - smoothing - 15 - Applies Gaussian smoothing based on scan count. The smoothing value is interpreted relative to the scan rate; for example, doubling the scan rate is roughly equivalent to halving the smoothing value by RT. Does not require reextraction.
  • Annotation RT - annRt - Expected RT of the annotation. This should usually be similar to the extraction RT.
  • FWHM - fwhm - 0.1 - Expected full width at half maximum of the peak. Integration cannot stop within this range.
  • Peak Search - peak_range - 0.2 - The ± RT search range used to find peaks, centered on the annotation RT.
  • Baseline Search - baseline_range - 0.3 - The ± RT search range used to find the base of the peak from the apex.
  • Peak Ranking - peak_rank - 1 - Method used to rank detected peaks. If multiple peaks are detected, this determines which peak is considered best.
    • Range - Special case where a time range is used, based on annotation RT and baseline search. Set Peak Start to 0 to enable this.
    • Tallest - 0 - Ranks peaks by apex height.
    • Estimated Area - 1 - Ranks peaks by estimated abundance based on the estimation line, described below.
    • Nearest RT - 2 - Ranks peaks by proximity to the annotation RT.
    • Left to Right - 3 - Ranks peaks from left to right based on RT.
    • Right to Left - 4 - Ranks peaks from right to left based on RT.
  • Peak Start - peak_start - 1 - Selects which ranked peak to start with. Use 0 for range-based integration.
  • Num Peaks - num_peaks - 1 - Number of peaks to include for integration, based on peak ranking and start number.
  • Spike Percent - spike_percent - 0.1 - Allows integration to ignore small spikes. In this example, spikes smaller than 10% of the running prominence may be ignored. This is also influenced by FWHM: the spike must be less than half the FWHM.

Despite the number of settings, integration is straightforward:

  1. The XIC is smoothed.
  2. An estimation line is generated to estimate the area at each point on the XIC. This is calculated as a rolling integration over the FWHM window.
  3. Local maxima are identified within the annotation RT ± peak search range.
  4. Peaks less than one-tenth of the estimated apex height are ignored.
  5. Peaks are ranked according to the selected peak ranking method.
  6. The first and last peaks are identified using the peak start, peak ranking, and number of peaks parameters.
  7. Integration-bound searching begins on both the left and right sides.
  8. On the left side, the bounds start at the leftmost selected peak apex minus half the FWHM, and the next local minimum is identified.
  9. The algorithm searches for the next lowest point that is both within half the FWHM of the current point and does not contain a spike, based on spike percent multiplied by the current peak prominence at the local minimum.
  10. This continues until the baseline search limit is reached.
  11. The right side is processed the same way.
  12. The Y-coordinates for peak integration dots are placed at the XIC intensity multiplied by baseline_percent.
  13. The integration dots are rolled toward each other until there are no negative points within the integrated region.

Status

TFE is designed to handle changes to both extraction and integration parameters. If these parameters are changed but the sample has not been reprocessed, status warnings will appear in several places. Status colors are listed below in ascending priority:

  • Green - Good. Extraction and integration are up to date with the current settings.
  • Light Yellow - Optional reextraction. The sample was extracted with different settings than the current settings, but the current integration is still valid. This usually occurs when the extraction range was larger than necessary and has been reduced.
  • Yellow - Reintegration required. The sample needs to be reintegrated. This usually happens when annotations.csv is changed or when settings are changed on the Annotations page. You can use Reintegrate All on the Samples page to update everything.
  • Light Red - Reextraction required. The old integration values are likely no longer valid. This can happen when extraction parameters are changed, usually PPM or MZ. It can also happen if Automatically Reextract is set to Never on the Settings page and Apply to All is clicked in the popup chart.
  • Bright Red - Incompatible settings. The extraction parameters are not valid for the integration parameters. This occurs when the XIC bounds, defined as RT ± extraction range, do not fully contain the integration bounds, defined as annotation RT ± (peak search + baseline search).

Status warnings appear in the following locations:

  • Samples Page - The table shows the highest-priority status for each sample. For example, light red means at least one annotation in that sample must be reextracted.
  • Peaks-Annotation / Peaks-Sample - The main table displays the current status.
  • XIC Chart - The chart background changes to the status color.

Library, Non-Library, Manual, and Cleared

Most annotations use the values in annotations.csv, which are referred to as the library settings. However, individual integrations can also use their own settings.

  • Library - Uses the default settings from annotations.csv.
  • Non-Library - Uses integration-specific parameters, usually set with the Apply to One button in the Popup Chart or by Alt-clicking the XIC chart.
  • Manual - The integration was adjusted manually.
  • Cleared - A special manual case where the peak is removed. Use the black X in the XIC chart to clear it. Use the red circle that appears to revert it to the library annotation.

Clicking Apply to All will not overwrite manual or non-library integrations. These integrations can be reverted by clicking the X and then the circle, or by using the Revert button in the Peaks-Annotation page toolbar.