Updates mostly aimed at increasing the responsiveness of ElasticSearch query resolving, plus an effort at abstracting away from MSK Spectrum specific tools.
We're also deeming this our official version 1 of Mira!
-
Input files now assume tsv/mtx/json files in one directory.
This should be easier to read into memory and MUCH faster to load. In addition, the loader no longer assumes any connection to an outside database - just give it the directory path with all the relevant files and it's good to go! -
Elasticsearch schema changes
We've now nested gene records inside their respective cell records, eliminating the need to do join-like queries when we add filters. This makes the graphQL resolvers much simpler, and queries should now return data to the front end faster. -
Marker gene flexibility
You can now specify different marker genes for each dashboard loaded. At this point, we still assume that each dashboard has exactly one marker gene table, but we're aiming to have this as an optional table in the future. There are still some bugs to work out, so for now we've removed the counts from the CellAssign table. -
Aesthetic changes
Quite a number of aesthetic changes, including replacing the dropdown of dashboard type with a tabbed view. We've also changed the font and several of the theme colours :)