How are the MS/MS matching score and similarity score calculated?

MetaboAnalyst supports MS/MS pattern matching-based compound identification since version 6.0, introducing two key scores for result evaluation: the matching score and the similarity score. The following provides an explanation of these two scores:

  1. Similarity Score: MetaboAnalyst employs the dot-product or spectral entropy method for its calculation. This score exclusively reflects the pattern similarity between the query spectrum and the reference spectrum.

  2. Matching Score: This score offers a comprehensive evaluation, taking into account MS/MS pattern similarity, m/z deviation (calculated by assessing differences between experimental and reference values based on the Gaussian distribution of m/z error), and the ratio of matched fragments.

In summary, while the similarity score concentrates solely on MS/MS pattern comparison, the matching score provides a more thorough evaluation of results by considering additional factors such as m/z deviation and the ratio of matched fragments.

2 Likes

Hi Zhiqiang,

Thank you for the helpful explanation on this topic. I have a quick question regarding the MetaboAnalystR/OptiLCMS local version. Could you confirm whether the matching score option is enabled for the local version? I wasn’t able to find any documentation on it.

It appears that the similarity score and dot product are used by default for the local version. Any clues you could share would be greatly appreciated.