civic actionability score revisit

[ahwagner]

I was just reviewing the code for CIViC actionability scores, and it looks like it relies entirely on evidence level and rating... but doesn't account for conflicting evidence at all. In fact, a hypothetical variant with 50 equivalent evidence items in direct conflict (e.g. "supports resistance to drug X in SCLC" vs. "supports sensitivity to drug X in SCLC") would have 2x the actionability score of a variant with 50 equivalent evidence items all in the same direction. In fact, the notion of evidence type, clinical significance, evidence direction, disease and (if applicable) drug context... none of it is accounted for.

It's clear to me after review what this is, and it's exactly as described in the help documentation, but I can't help but feel calling it an "actionability" score is a bit misleading--wouldn't something like an "evidence score" be more appropriate here?

This doesn't need to be resolved now or anything--I'm just making a note here that we should maybe revisit this before the 2.0 paper, where we will presumably summarize all the new stuff since the 1.0 release (which, I believe, includes actionability scores).