Performance boost

For an input topic, Génie ranks all genes from a species using their related biomedical abstracts. For a gene, if a significant number of gene-related abstracts are found associated to the input topic by the Génie's text classifier then the gene will have a better rank. Similarly, Alkemio ranks chemicals using their related biomedical abstracts. Notably, the text classifier uses only words in abstracts for its computations. But now, if the input topic is defined by a user as a single MeSH term, Génie or Alkemio will first search it in abstracts MeSH annotations. If it can find the term, the abstract is automatically considered positive (p-value set to 0), otherwise a p-value is computed using words. Display of top 10 abstracts on the output page has been updated. Abstracts having a p-value=0 are displayed using ♥, abstracts having a p-value for high precision are displayed using ⋄, abstracts having a p-value for good precision are displayed using ♠, and other abstracts having a p-value below the input cutoff are displayed using ♣.

Comments are closed.