Hi there ! I'm loking for a tools that is able calculate a similarity score for each term of 2 list name and give me the top 10 similarity score he found. For exemple List 1 : (-)-epigallocatechin (+)-catechin (pyro)catechol sulfate 3',4'-Dimethoxyphenylacetic acid List 2 : 3',3'-Dimethoxy-phenylacetic acid catechin (epi)gallocatechin catechol sulfate Results (-)-epigallocatechin VS (epi)gallocatechin - Score = 0.9 (very similar) (-)-epigallocatechin VS (+)-catechin - Score = 0.5 (+)-catechin VS catechin - Score = 0.9 etc etc ... Thanks a lot for your great help.
In case you have ChEBI identifiers:
https://github.com/lasigeBioTM/DiShIn
to obtain ChEBI identifiers from text:
https://github.com/lasigeBioTM/MER
In order to determine the similarity of arbitrary terms, I would transform them into their sets (alternatively Bag-Of-Words BoW) and would use Jaccard similarity on the two sets (resp. an weighted derivation for BoW).
However, better suited would be to transform them into a set of their syllables, that would respect the domain language better. But I don't know whether that can be done for chemical compounds or molekules easily.
Hope that helps for first attempt.
http://norvig.com/spell-correct.html
Although this is not a tool in the common sense, the link above contains working code and good explanation of how to use edit distance to calculate similarity between words. With some.modification, you can create the tool you want.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining