Following alignment, the resulting peak list may contain missing peaks as a product of a deficient peak detection or a mistake in the alignment of different peak lists.
The fact that one peak is missing after the alignment does not imply that the peak does not exits. In most cases it is present but was undetected by the previous algorithms.
This algorithm fills the gaps in the peak list when it is possible according with the parameters defined by the user. The most crucial parameters are "m/z tolerance" and "RT tolerance" which define the window where the algorithm should find the new peak. It is centered in the m/z average and retention time average of the source peak list. Once the best candidate is found inside the window, its intensity and its shape in RT direction is also checked.
It can also add a previous correction of the retention time in the case it is needed. It will change the position of the defined window depending on the prediction of a RT model created using all the already aligned peaks in each pair.
When RT correction is applied, the algorithm is divided in two main steps. In the first step, one random sample is taken from the multiple peak list and is used as a master list. All the gaps of this master list are filled using all the others samples. For each pair of samples the algorithm creates a model of the retention time. In the second step the master list is used to fill the gaps of the rest of the samples, creating also a retention time model for each pair (as is is showed in the figure below).
New peak list showing the filled peaks with a yellow mark.