Biomedical Engineering Reference
In-Depth Information
40
35
30
25
20
15
10
5
0
0
5
10
15
20
25
30
35
40
Probability state modeling S-phase
FIGURE 14.6
Comparison with conventional least squares modeling. One hundred and fifty
DNA listmode files with 10,000 events and 3%CV were generated with %S-values ranging
between 10% and 35% and were analyzed by the probability state modeling system (x-axis).
The same data were read and analyzed by a cell cycle modeling package (Zz13) and the
%S-phases were compared. The results indicate similar accuracy between these approaches.
a general way. It largely eliminates the error and subjectivity associated with creating
analysis regions or quadstats. These types of gate-based constructs have hard
boundaries that immediately create false positives and negatives. Hierarchical gating
compounds these errors since the output of one gate is the input to another. Probability
state modeling boundaries can be considered soft and malleable and can change as
more parameters are added to the model. We will see plenty of examples of this a little
later in the chapter.
14.5 TIME-DEPENDENT PROGRESSION
So far, I have been using cell age and cell cycle as an example to help you understand
how probability state models work. It is time to take the training wheels off and
consider the real complexity of many of our samples. I used cell age as an example of
a time-dependent axis, but the reality is that this concept is much more general than
cell age. The proper way of abstracting cell age is to consider it as an example of
some time-dependent progression. This abstraction allows us to look at other
progressions, such as cell differentiation and maturation, in the same way as we
have with cell cycle progression. It really does not matter to us what progression we
are interested in since we can always map these time-dependent progressions to our
probability state axes.
Search WWH ::
Custom Search