This method aligns detected peaks in different samples through a match score. This score is calculated based on the mass and retention time of each peak and ranges of tolerance stipulated in the parameters setup dialog.
The peak alignment algorithm uses:
1. a master list of peaks (L) against which every new sample (Sj) will be matched. When aligning peaks from multiple samples, the master list is initially set to the first sample, and subsequently it will be the combination of samples aligned thus far, with the samples as the columns and the matching peaks as the rows.
2. for every row i in L, a two-dimensional window (where the window size is selected by the user), called Alignment window (AW) defining the ranges of m/z and RT and centered around the average of m/z and RT of all the individual peaks in the row; and
3. a score function to compute the similarity of peaks between L and the new
sample Sj inside the alignment window. The score function computes the similarity based on
the similarities in m/z, retention time, and optionally on identification, and isotope patterns
between the peaks to be matched. The score is calculated as follows:
score = ((1 - MZdifference / MZtolerance) × MZweight) + ((1 - RTdifference / RTtolerance) × RTweight)
The algorithm works as follows. It iterates through the rows of L, and for each row, it looks for peaks within the alignment window in Sj that has to be aligned with L. A score is calculated for each possible match and the pair getting the best score will be aligned.
New aligned peak list showing peaks from 3 different samples.