Bug in detection p-value

[Jari Häkkinen]

This bug affects all previous versions of the detection p-value plug-in. The error is small and has minor impact on calculated p-values.

The MAD outlier part of the algorithm sometimes excludes small intensity negative control probes from the detection p-value calculation. This case was not considered prior to this fix. This omission decreased the p-value with a small number and consequently the smallest p-value could become less than zero and p-value 1 could not be reached.

The resolution of this ticket solved the above problem and changes the border for zero p-value such that no probe with intensity equal or less than the largest negative control can achieve a zero p-value. Note, there may be negative control probes with larger intensities than the ones used in the detection p-value calculation (MAD outliers).

[Jari Häkkinen]

(In [1232]) Fixes #273