Figure 6

Noise test for two human lung samples replicated previously described. During a test, a certain number of perfect matches were randomly selected, and their intensities were multiplied by ten. The computations were otherwise done in the same way as in Figure 2. The MAS4 and MOID calculations are in blue and red filled circles, respectively. The error bars represent the widths of intensity ranges where 80 percent of the fold change data fall. Both methods are robust enough to stand for two noisy probes; MOID is slightly in favor for larger noise.