speed up results for large pathways (cache, filter)?

[nathandunn]

It is several MB's of data vs several K for the default. The default app does this via multiple smaller calls, but also for calls for ONLY the samples needed. So the solutions are:

1. hand search / sort parameters in
2. sort and then gene sets on the server for activity
3. cache results for each side?

---

Easiest might be:

• [ ] run tests over all the different types
• [ ] trigger cache of samples on job upload (for empty / all samples)
• [ ] trigger new cache when samples are uploaded
• [ ] generateResult could use the filter to return once calculated (or a follow-up method)

---

Note that we can cache the meanmap score over all samples for all of them, especially, if we can limit the number of gene sets, etc. . . . however, we will need to calculate from scratch, otherwise.

[nathandunn]

Probably the best strategy would be that on analysis, for each cohort we automatically store the returned results for each cohort with a max filter of 200 or so, one for each sorting preference.

[nathandunn]

this might be difficult as the mean map is necessary to calculate samples as one needs to re-order samples based on what is passed in.