Biomedical Engineering Reference
In-Depth Information
14.8.3 Studying Perturbation Effects of Drugs
Progression-specific scaffolds will also undoubtedly play an important role in the
pharmaceutical industry. Although there are no studies to date using probability state
models to examine how drugs perturb specific aspects of in vivo or in vitro cell
populations, there are some characteristics of this system that make it quite attractive.
Since there are a fixed number of states in a probability state model, the number of
cells necessary to achieve statistical significance does not increase geometrically with
increasing number of parameters. Partitioning schemes using parameters as their
boundaries will always have this unfortunate side effect. The other advantage is that
any change in population percentage or expression pattern will perturb the state
frequencies. The reduced chi-square (RCS) is typically used to measure this degree of
nonuniformity and since RCS can be directly related to a probability value, sig-
nificance can be estimated from a single sample as opposed to a set of replicates.
14.9 OPTIMAL PANEL DESIGN
In the future, panel design will evolve to carefully constructing tubes with common
markers that defined one or more progression-specific scaffolds. These scaffolds
will play an increasingly important role in diagnostic cytometry. These scaffolds will
allow probability state models to assess all the normal populations and identify
specific abnormal populations or abnormal shifts in populations.
14.9.1 Avoiding the Dimensionality Barrier
In order to get some sense of how different probability state modeling is from
conventional cytometry, you should realize that it would take 45 two-parameter
histograms (10
9/2) convey the same amount of correlation as shown in
Figure 14.11. Even if we only select those parameters necessary for the scaffold,
a total of 10 histograms would be necessary (5
4/2). Thus, the dimensionality
barrier that has been limiting cytometry for the past few years is largely eliminated by
these parametric plots. As soon as we decided to look at one measured parameter
versus another, we started down this road that had a seemingly insurmountable barrier
at the end.
14.10 THE MUSIC ANALOGY
We can use another analogy to better make this point. Imagine if musical scores
were written with one musician
snotes
on the other. Could the musicians play their notes with this type of representation?
What happens when there are 10 musicians or 100? I began this chapter showing
some simple parametric equations involving the common variable, t. These simple
equations can create surprisingly complex figures when the data are graphed with
s notes on one axis and another musician
Search WWH ::
Custom Search