Figure 27. Scatter plot of the ion intensities at m/z 32 versus m/z 101, showing absence of correlation with the disease as well as between both ion intensity distributions. Note high value for sample C (female control) at m/z 101.

m/z 32 appears to be dominated by strong individual differences between all three controls (compare this finding with the pronounced differences in clinical data for these samples in Table 4), whereas m/z 101 shows an exceptionally high value for the only female sample (control C). Additional analyses of different samples will be required in order to confirm the significance of these observations.

Similar two-dimensional plots could be constructed, of course, for many other combinations of masses with high "characteristicity" values. However, a far more comprehensive approach is to use multivariate statistical analysis techniques which enable a comparison of intensity distributions of a much larger collection of masses to be made simultaneously. Also, this approach avoids the danger of misjudging the significance of seemingly non-random distributions when examining isolated features for an expected differentiation in intensity distributions. Obviously, a significance level of 1% (p <_ 0.01) does not prove differentiation if only observed a few times among a hundred or so different features inspected.

0 0

Post a comment