First Pass Annotation

The first pass annotation is the first step in the annotation process and is based on several filters and algorithm to get the best possible result. The most important filter of this is the retention index filter and everything else is needed to refine the results.

• retention index filter
• base peak filter
• forbidden ion filter
• sifter filter 
  • similarity
  • purity
  • signal noise
• ion ratio filter

The first step is to sort all peaks by there signal noise. This makes sure that we work on the tallest peaks first and than on the little peaks.

The next step is to go over all the peaks and look for a match in the given retention index window range, which can be defined for each bin, but is set to a default of +/- 2000 RI units.

If we found a match using this we check that the unique ion is at least X% of the base peak and if this is the case we move on to the next filter.

The sifter filter is now applied to the current mass spec which checks that the 3 properties of it (purity/signal noise/similarity) are in the expected ranges and if this is the case we add this peak as possible annotation.

To give the best possible result we do a forward matching, reverse matching and run at the end of this process the double annotation detection, if we had double annotations. A forward matching is the comparison of all bins against all spectra and the reverse matching is the comparison of all spectra against all Bins.

In case of double annotations there are now several algorithms applied, depending off the size of the peaks.

If both peaks have a small signal noise we know that they have most likely not a major retention index shift, since this only happens with overloaded peaks. In this case we try to take the peak which is closer to the bin, if the other peak has not a significant higher similarity. In case of a significant higher similarity we take this peak and return the other peak to the list of available peaks for annotations. Otherwise we use the closer peak.

Now if both peaks are taller than the defined signal noise level, than we use the one with the higher intensity, or if both have the same intensity we use the default small peak algorithm. If one peak is bigger than the defined signal noise level and the other one is smaller than we accept the one with the highest signal noise and return the other peak to the list of available peaks.

Once we annotated all mass specs, we try to generate bins from it in the next step, for this they need to have a certain purity and signal noise. And need to be found in X% of a calss. This is done to avoid that the database is corrupted with a lot of artifacts, which have no real meaning to science.

The results of this annotation process are colored in a light green in the exported XLS file.