1
$\begingroup$

I have a set of cell line pairs and I want to know to what extent the pairs are similar based on their ontologies. The problem I have is that I have found a Python library called Fastsemsim, but it has many similarity measurements. Or to be exact, there are actually many ontology similarity measurements. And I want to know which similarity measurement is best suited for my case (cell line ontologies similarities). There might be a couple of good ones that are good for cells ontologies. These are a set of similarity measurements that are available in Fastsemsim.

Semantic Similarity

  • SemSim.TermSemSim
  • SemSim.ObjSemSim
  • SemSim.ObjSetSemSim
  • SemSim.SetSemSim
  • Specific Semantic Similarity measures
  • SemSim.MixSemSim
  • SemSim.avgSemSim
  • SemSim.maxSemSim
  • SemSim.BMASemSim
  • SemSim.SemSimUtils

Source: https://fastsemsim.readthedocs.io/en/latest/library.html

$\endgroup$
  • $\begingroup$ Welcome to the site. Which is your question. What is the best semantic similarity? Or something else (do you want to compare the cell lines with non-semantic similarities measures)? Do you have other information besides cell lines ontologies (not sure were do you take those from or in which format though)? $\endgroup$ – llrs Jul 22 at 8:03
  • $\begingroup$ The format is OBO but that is not related to the question. My question is which similarity measure (listed) is best or does better at finding semantic similarities between cell lines $\endgroup$ – Yousef Jul 22 at 11:38
1
$\begingroup$

About the metric you use, there are several reviews on the topic, if I recall well I think that the information content based are the best ones, of those, I only recognized the JiangConrathSemSim listed on the webpage.

However, if you want to use a semantic similarity measure I would use the BMA method, which is more comprehensive than other methods. But the BMA method is only a way to aggregate several metrics of ontologies into a single numeric value. I am not confident that FastSemSim implements correctly those methods.

Note, that according to the license, you should contact the authors before using it:

You must contact us before using it or its results or any application/tool based on it in any published work.


I am not sure that the cell lines fit in a ontolgogy. Because ontologies are defined as directed acyclic graph (DAG) and there are some cycles that happens in cells like the induced pluripotent stem cells show. However they might be right if you just compare the cell lines and not the state of the cells on those lines (use it only if you want to compare the lineage of your cells lines).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.