Python for Bioinformatics

27 Applications with Expression Arrays

The M versus A plot from the previous chapter displays additional problems in the data. At lower intensities, there is a tendency for the second channel to be more expressed than the red channel. This is seen in the data, which has a general slant rather than a horizontal distribution.

27.1 LOESS Normalization

The LOESS normalization procedure, which was discussed in Section 5.3, is a simple method of removing this bias. The argument as to the validity of removing the bias is beyond the scope of this text. This method basically separates the data along the horizontal axis. For example, the data can be separated into five groups, with 20% of the data points in each group. Obviously, the groups will not have ...