#258 Percolator performs poorly in conjunction with exact p-values

[William S Noble]

In the performance test, the line for Tide with exact p-values followed by Percolator is far below the line for Tide with exact p-values alone. This does not happen for q-ranker, nor does it happen for Comet or Tide without exact p-values.

[Jeff Howbert]

When I use Percolator with exact p-values, I convert to -log10(p-value) first. I don't think it's reasonable to expect a linear SVM to learn that the difference between 10e-12 and 10e-8 is the same as the difference between 10e-7 and 10e-3.

Jeff

[William S Noble]

I agree with this assessment. Sean or Kaipo, could you comment on how easy/hard it would be to have crux percolator automatically do a log transform on the "exact p-value" feature prior to passing it to Percolator for analysis?
Bill