Spectral Pre-processing

From MicrobeMS Wiki
Revision as of 10:26, 18 March 2023 by Laschp (talk | contribs) (Created page with " == Introduction == Spectral data analysis is a challenging task of the MALDI-TOF MS-based workflow for microbial identification. Considering that a MALDI-TOF mass spectrum is a complex signal which usually consists of hundreds of peaks and certain level of noise, adequate spectral pre-processing is required prior to peak detection and classification analysis.<br> The quality of microbial mass spectra should be assessed visually immediately after data acquisition with r...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Introduction

Spectral data analysis is a challenging task of the MALDI-TOF MS-based workflow for microbial identification. Considering that a MALDI-TOF mass spectrum is a complex signal which usually consists of hundreds of peaks and certain level of noise, adequate spectral pre-processing is required prior to peak detection and classification analysis.
The quality of microbial mass spectra should be assessed visually immediately after data acquisition with regard to the following criteria: first and foremost, the signal-to-noise ratio (SNR) and the presence of a sufficient number of mass peaks needs to be evaluated. Further quality criteria are a relatively flat shape of the spectral baseline and the absence of interfering, or confounding, mass peaks from plasticizers or other synthetic polymer additives. Outliers, i.e. spectra failing to meet one or more of the quality requirements should be not accepted for multivariate classification analysis and thus excluded from further analyses.

Smoothing

This function is used to smooth mass spectra using the Savitzky-Golay smoothing filter. Smoothing has a mostly cosmetic effect on the spectra, reducing the noise at the expense of lowering the resolution of mass peaks. Possible values for smoothing points are 3 to 75 (see popupmenu # smooth pts in the PREPROCESS tab of the main figure of MicrobeMS. If the checkbox org. spectra is checked smoothing will create pre-processed spectra from original spectra. Note that existing pre-processed spectra will be overwritten without warning in this case. Press the smooth button to smooth selected mass spectra.

Details of the Savitzky-Golay algorithm can be found in the literature:

    A. Savitzky and M. Golay. Smoothing and Differentiation of Data by Simplified Least Squares Procedures. 
    Anal. Chem. 1964 Vol 36(8):1627.

The parameters used for smoothing are stored within the program workspace and are accessible through the FILE INFO tab (press button edit of the FILE INFO tab, or select edit header info from the Edit pulldown menu).

Baseline correction

The function divides a spectrum in segments, or intervals, for each of which a minimum MS intensity values is determined. These values are in the following used to generate a baseline correction curve (by shape-preserving piecewise cubic interpolation). For baseline correction the curves are subtracted from the original spectra. To perform baseline correction select first the spectra to be corrected from the listbox in the upper right corner (Screenshot of MicrobeMS). Select then the number of intervals from the popup menu # of intervals in the VIEW tab. Allowed values for the number of intervals (niv) are 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23, 25, 27, 29, 31, 35, 45, 55, 65 and 75. Subsequently press the button baseline (VIEW tab). If the checkbox org. spectra is checked baseline correction will create pre-processed spectra from original spectra. Note that existing pre-processed spectra will be overwritten without warning in this case. To perform baseline correction from existing pre-processed spectra uncheck the checkbox org. spectra.

Baseline correction should be repeated from the original spectra in the event of negative intensities after baseline correction. For this purpose the checkbox org. spectra should be checked and the parameter # of intervals should be reduced to 60 or less. Again, existing pre-processed spectra will be overwritten without warning.

The parameters used for baseline correction are stored within the program workspace and are accessible through the FILE INFO tab (press button edit of the FILE INFO tab, or select edit header info from the Edit pulldown menu).

Normalization

This function normalizes mass spectra. Normalization is carried out in the following way: spectra are first mean-centered, i.e. the average value of the intensities is calculated and subtracted from the spectrum. Then, the spectrum is scaled such, that its standard deviation equals 1000. To obtain a mass spectrum with a baseline at intensities around zero, the most frequent spectral intensity value is determined which is finally subtracted from the mean-centered and intensity-scaled spectrum.
Normalization does not require parameters. The normalization function is available from the Preprocessing pulldown menu or via the button normalize in the PREPROCESS tab.

Cut

Screenshot of the window "cut mass spectra"

Cutting mass spectra is useful to narrow the mass range of the spectra. In case of large data sets this may be useful to free some memory before memory-consuming calculations are carried out. Define the mass range to be kept [m/z], then press the cut button to start the function. Leave the function by pressing the button cancel.
When the checkbox org. spectra is checked (see the VIEW tab in main user interface of MicrobeMS) cut overwrites existing pre-processed spectra without warning. If pre-processed spectra are not available cut will create pre-processed spectra from original mass spectra. Check the checkbox replace also original data to cut original spectra. Note that cutted original spectra cannot be returned to the original state (see also pre-processing function clear pre-processing).

The parameters used by the cut function are stored within the program workspace and are accessible through the FILE INFO tab (press button edit of the FILE INFO tab, or select edit header info from the Edit pulldown menu).

Auto-calibration

Screenshot of the window autocalibration

Performs a linear re-calibration of the mass spectra. Note that methods other than linear are not available. Auto-calibration overwrites existing pre-processed spectra and modifies also existing peak tables. If pre-processed spectra are not available auto-calibration will create pre-processed spectra from the original mass spectra, i.e. the original spectra are not modified by this function.
The spectral pre-processing procedure of auto-calibration is based on the utilization of mass peaks of the analyte, i.e. it requires the knowledge of the precise peak positions of at least two different sample peaks per mass spectrum. To perform auto-calibration, it is necessary to produce peak tables from adequately pre-processed mass spectra. Mark then spectra to be auto-calibrated and select autocalibrate from the Preprocessing pulldown menu.
This will open a dialog box entitled autocalibration of mass spectra. Indicate the precise m/z positions of at least 2 and not more than 30 mass peaks and define the estimated calibration error from the popupmenu allowed mass tolerence (ppm). Select larger values if you are unsure on the actual calibration error. The pseudo-gel view may be helpful to define the positions of peaks suitable for auto-calibration. Press calibrate when finished. IMPORTANT: At least two of the peaks used for autocalibration should have counterparts in the peak tables of the uncalibrated spectra. Its worth to check the output of the command line window when auto calibration has finished.
The parameters used for auto-calibration are stored within the program workspace and are accessible through the FILE INFO tab (press button edit of the FILE INFO tab, or select edit header info from the Edit pulldown menu).

Reduce resolution

Screenshot of the window reduce resolution

This function allows to reduce the effective spectral resolution of the mass spectra by a factor of choice. In case of large data sets this may be useful to free some memory before memory-consuming calculations are carried out. Select the data reduction factor from the reduction factor popupmenu (3-21), then press reduce to start the procedure. Leave the function by pressing the button cancel. When the checkbox org. spectra is checked (see tab VIEW/main window) reduce resolution overwrites existing pre-processed spectra without warning. If pre-processed spectra are not available reduce resolution will create pre-processed spectra from original mass spectra. Check the checkbox replace also original data to modify also the original spectra. Note that modified original spectra cannot be returned to the original state (see also pre-processing function clear pre-processing).

Clear pre-processing

This option can be used to delete selected pre-processed mass spectra from the MicrobeMS workspace. Note that original spectra processed by the functions cut and reduce resolution cannot be returned to their original state. The function clear pre-processing is available from the Pre-processing pulldown menu.