Feature idea for lists: calculate improbable overlaps

[yochannah]

Snazzy List enhancement idea from Gos:

List view:

• automatically calculate whether there are improbably overlaps with any of
  (potentially many) personal lists, or public lists.

  • make it very easy from that point to extract the common things into a new list
    i.e. in one click derive the set of your favourite genes that are also essential.

• The above, when in list view, could operate on all public and your private lists.
  However it could be very useful if enrichments are done automatically in the background
  whenever you run a query. i.e. you run a template that happens to return genes as one
  of the columns. In the background it checks and finds a) improbable overlap with
  cancer genes; b) improbably overlap with one of your private lists; c) GO term
  enrichment. Maybe it would take too long to run but it could be cool.

[yochannah]

see also:
https://en.wikipedia.org/wiki/Jaccard_index
https://en.wikipedia.org/wiki/MinHash

[julie-sullivan]

@nupurgunwant is interested in working on these types of problems!

[julie-sullivan]

We use the hypergeometic distribution for enrichment now: https://intermine.readthedocs.io/en/latest/embedding/list-widgets/enrichment-widgets/

[julie-sullivan]

Also depends on the mine and / or the user having interesting lists already.

Running enrichment in the background for each query sounds expensive.

[yochannah]

Should this ticket be in BlueGenes? I'm thinking we could test out GitHub's new snazzy migrate-a-ticket feature to send it to InterMine, maybe.

…

On Fri, 2 Nov 2018 at 09:30, Julie Sullivan ***@***.***> wrote: Also depends on the mine and / or the user having interesting lists already. Running enrichment in the background for each query sounds expensive. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#259 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AI14ji9QCIdMhm3oTx7-t-lA6ELLUIBZks5urBCygaJpZM4UkT42> .

[julie-sullivan]

should really be on a google doc wishlist somewhere, until we decide to implement it.

[yochannah]

The endpoints for this will be released soon - see intermine/intermine#2029

We'll need to conditionally add this to the list page, once intermine 4.0.0 is released, and check for the endpoint existence or correct im version before displaying it.