Closed asizemore closed 11 months ago
maybe any time we return pvalues, we should return the pvalue floor used as well? then you dont have to have a separate variable frontend?
for the adjusted pvalue, im not sure i understand.. is assigning one based on the floor any worse than assigning a floor in the first place? the most accurate way to represent these floored values for both the pvalue and adj pvalue would id think be to do something like '<= pvalue' and '<= adj pvalue'
comments addressed! Pvalue floor is now a prop, the pvaluefloor and adj pvalue floors are returned with the data, and we do a real calculation for the adj pvalue floor.
Please give a close eye to my adj pvalue calculation!
@d-callan ready for another review!
@d-callan both comments resolved in the latest commit
Resolves #51
This PR adds a variable
P_VALUE_FLOOR
to the package. We can use this floor in differential abundance now, and in correlation later. In short, we replace any pvalue smaller thanP_VALUE_FLOOR
withP_VALUE_FLOOR
. The frontend will also get a matching variable with a comment to ensure the two stay linked.The one wrench in this plan is the adjusted p values. Since we can't just assign one based on the floor, and using the original adjusted p-value doesnt make sense since then there would be two points at
P_VALUE_FLOOR
with different adjusted p values, I've assigned those modified points to have an adjusted p value of NA. Since the point is to explore, i think a tiny p value should suggest to the user that the point is worth exploring further and on their own. It's not my favorite solution, so other ideas are welcome!