@zorian15 Some changes to how the epitope harmonization is done. Can you review this?
Previously the epitope harmonization assumed epitopes were integer-numbered sequential, and then adjusted the params. The method didn't work if they had string rather than integer names, such as epitope_1, epitope_2, etc. This is bad, because in practice we may be naming them.
Now we simply require the epitopes to have the same names, which can be anything. The epitope labels are then just adjusted, rather than re-ordering the params. So the key thing is that output matches by epitope name, not by order in the self.epitopes attribute. This is better because we will work with data frames with the epitope names as labels.
The correlation data frames are made available to be returned and viewed in public methods.
And you shouldn't really call it a "bug", Zorian. Your code worked as implemented, but I just thought it should be extended to cases where epitope names aren't default of sequential numerical.
@zorian15 Some changes to how the epitope harmonization is done. Can you review this?
Previously the epitope harmonization assumed epitopes were integer-numbered sequential, and then adjusted the params. The method didn't work if they had string rather than integer names, such as
epitope_1
,epitope_2
, etc. This is bad, because in practice we may be naming them.Now we simply require the epitopes to have the same names, which can be anything. The epitope labels are then just adjusted, rather than re-ordering the params. So the key thing is that output matches by epitope name, not by order in the
self.epitopes
attribute. This is better because we will work with data frames with the epitope names as labels.The correlation data frames are made available to be returned and viewed in public methods.
The code was streamlined a bit.