Open sehilyi opened 2 years ago
One thing to note is that TABIX supports separate types of indicies: traditional FAI and CSI. Since we make the index path explicit, for these file types, I think that we can determine which index is used from the index URL (i.e., .tbi
vs .csi
), but we will probably want to support both.
Another thing we could learn from jbrowse folks is coming up with a shared abstraction for retrieving "rows" (or features) or these abstract data formats:
That way rather than implementing N data-fetchers which wrap tabix, we can perhaps have a shared underlying object which performs fetching & parsing given a TabixIndexFile
and some parser
. So the adapters just need to initialize the TabixIndexedFile
and parser
.
Given that BED files are very common, we need to support Tabixed BED as well.
Reference
Closely related issues