Skip to content

Sample Scores

Description

  • Diversity score: summarizes the chemistry that is found in this sample, compared to all other samples. Here, spectral similarity networks are taken as proxy for chemical families. For each spectral similarity networking algorithm, the number of detected networks for this sample is divided by the total number of networks, with the quotient being the score. Only the highest quotient across all networking algorithms is retained. The higher the score, the more different chemistry can be observed in the sample. Diversity scores strongly depend of networking parameters and must only be considered for a single fermo analysis run.
  • Specificity score: summarizes the chemistry that is specific to this sample, compared to the other samples. For each networking algorithm, the number of detected networks for this sample is subtracted by the number of networks that were also observed in other samples and divided by the total number of networks, with the quotient being the score. Again, only the highest quotient across all networking algorithms is retained. The higher the score, the more chemistry is specific for the individual sample. Diversity scores strongly depend of networking parameters and must only be considered for a single fermo analysis run.
  • Mean novelty score: The mean novelty scores of all features in this sample.