Span based search

[alancleary]

In the search view, the user should be able to provide a genomic interval for a chromosome, perhaps on the current query chromosome, for which to perform a new search. Ideally this would just be a matter of expanding the UI to allow the user to interact with the existing "Nearest Gene" service.

[adf-ncgr]

we might want to consider making this independent of the nearest gene service; I could imagine, for example, selecting a macrosynteny block and have that determine the span to be searched. And although
getting ridiculously long spans is a potential concern, it would certainly be simple to form a query track with a span and nothing else. perhaps numNeighbors could be used if present and if not present then the entire span would be used. not sure a good idea, just thinking that in many contexts when a span is supplied, the user will get confused if a magical transform to a subspan is effected for their comfort and safety (I suppose they could be warned)

[adf-ncgr]

another note to future selves (mostly future Alan) when it comes to implementing this one; there seems to be no reason in principle why this couldn't be extended to a multispan mode so that we could complete our set of options with pleasing symmetry:

1. entering a gene name and finding syntenic contexts
2. entering a set of gene names and getting the aligned syntenic groupings among them
3. entering a single span and finding syntenic contexts
4. entering a set of spans and getting the aligned syntenic subgroupings among them

One could imagine a set of spans being determined by something like a query (not the responsibility of GCV) for the set of regions associated with a trait across a number of species (ie QTLs or GWAS-defined areas).

Also note that 4) would be a nice way of getting from search mode into multi-alignment mode; although possible now using the families trick it suffers from the problem that any track having a deletion of the selected gene family will be left behind....

[adf-ncgr]

@sammyjava can you remind me of your use case for this so we can take into account when setting priorities for our next round of development?

[sammyjava]

Just the CSw8 example: I've got a genomic range from QTL markers, and I'd like to simply enter the range between the first and last to get the corresponding range on the context viewer.

[adf-ncgr]

FWIW, Nevin Young also asked for this functionality...

[adf-ncgr]

And nearly a year later, Nevin still yearns for it; let's make it happen before Ardmore!

[alancleary]

Implemented a span-to-context service in commit: f097efb

Implemented URL span-based searching from the client in commit: c8c900b

Turn a span on a chromosome into a GCV search via the client URL as follows: /search/<source>/<chromosome>/<span>. For example: /search/lis/phavu.Chr02/13142457-13496658

[alancleary]

Commits 1946d80 and 1477d0b added span-based search support to the search widget in the client.

[alancleary]

Commit a4b6c6d updated the Search component to alert the user when a span request is being processed and when it fails. The SpanSearch guard now immediately directs users to the search view and doesn't redirect if the span request fails.

Edit: This should be considered a temporary solution until we settle on how to handle fuzzy searches, as discussed in issue #94.

[alancleary]

Added support for spans defined using the ".." syntax in commit 67ed8ae.