median

This method assumes that the red and green intensities are related by a constant factor, i.e $ R=k \cdot G$. After the normalisation the centre of the log ratio distribution is therefore shifted to zero,

$\displaystyle M=M-c.$    

A common choice for $ c$ is the median of the M values for a particular gene set [7]. This procedure is not commonly used, since the red and green intensities often do not follow this relation [4].

Bjørn Kåre Alsberg 2006-04-06