Centroiding settings
If an experimental mass spectrum is supplied in the form of a profile, it must first be centroided before a search can be performed. Experimental spectrum type must be selected by opening either the ‘centroid spectrum’ or ‘profile spectrum’ tab. If ‘centroid spectrum’ is selected, no additional inputs are needed. If ‘profile spectrum’ is selected, parameters for the centroiding algorithm need to be provided. IsoGlyph has two built-in centroiding algorithms, which are implemented in pyOpenMS: HiRes and Iterative. HiRes is recommended for Orbitrap spectra, while Iterative is preferred for spectra from other types of instruments.
Centroid spectrum
Section titled “Centroid spectrum”If ‘centroid spectrum’ is selected, IsoGlyph will assume that the experimental spectrum is already centroided and will not apply a centroiding algorithm. No additional parameters need to be provided in this case.
Profile spectrum
Section titled “Profile spectrum”If ‘profile spectrum’ is selected, IsoGlyph will assume that the experimental spectrum is in the profile form and needs to be centroided. Additional parameters need to be provided and a centroiding algorithm needs to be selected in this case. A Savitzky-Golay filter is applied to reduce noise prior to centroiding.
S.G. frame length
Section titled “S.G. frame length”Savitzky-Golay filter frame length controls how aggressive the smoothing filter is. It must be a positive odd integer. It needs to match the feature size. A good rule of thumb is to set the frame length between 1 and 2 times the full width at half maximum of the specific peaks that need to be preserved.
S.G. polynomial order
Section titled “S.G. polynomial order”Savitzky-Golay polynomial order controls smoothing intensity and peak shape preservation. It must be a positive integer strictly less than the frame length. Low polynomial orders may distort peak shapes, while higher polynomial orders preserve narrow features better. A polynomial order too high can reintroduce noise.
Signal to noise ratio
Section titled “Signal to noise ratio”Signal to noise is a peak validity threshold. It is used for peak detection and peak boundary estimation by the centroiding algorithm. It must be a floating point number greater than or equal to zero.
Centroiding algorithm
Section titled “Centroiding algorithm”A centroiding algorithm can be selected by opening either the ‘HiRes’ or ‘Iterative’ tab. Then parameters for the selected centroiding algorithm need to be provided. If ‘HiRes’ is selected, then ‘peak separation’ needs to be provided. If ‘Iterative’ is selected, then ‘peak width’ and ‘spacing difference’ need to be provided.
Peak separation is the minimum m/z distance between peaks. Peaks closer than that distance will be merged. It is provided in atomic mass units (amu) and must be a floating point number greater than or equal to zero.
Peak width is the expected half-width of a peak. It is provided in atomic mass units (amu) and must be a floating point number greater than zero.
Spacing difference controls how large an m/z gap between adjacent data points may be while the algorithm continues extending a peak. It is unitless and is multiplied by the smallest local spacing. It must be a floating point number greater than or equal to zero.