Biology Reference
In-Depth Information
5.0 units of figure 4.14) corresponds to 1 tpm. This is the conversion
factor used below in comparing the relative noise levels of MPSS
and Affymetrix expression data.
Comparison of Relative Sensitivity of MPSS
and Affymetrix Experiments
When comparing noise levels between dissimilar experiments, it is
important to consider the possible contribution of “human factors” on
the observed fluctuations. For example, consider the two MPSS exper-
iments discussed in this chapter. The noise observed between biological
replicate measurements taken from two distinct RNA samples taken
from plated human breast cancer cells (replicates A and B in figure 4.7),
which is plotted as circles in figure 4.8c, is significantly higher than
the noise observed between biological replicates of RNA taken from
LPS-activated macrophage tissues (plotted in figure 4.11b). In fact,
the total noise observed between the macrophage tissue replicates is
comparable to that observed between replicate bead-loading measure-
ments in the breast cancer replicate experiments of figure 4.7. This
strongly suggests that human factors in the tissue preparation and RNA
extraction introduced significantly greater variability into the data sets
shown in figure 4.8 than they did into the data shown in figure 4.11.
In the following analysis, it is chosen to compare the replicate noise
observed in the MPSS experiments on RNA taken from the LPS-
activated macrophages with the replicate GeneChip measurements in
which RNA was also extracted from different cultures, that is, experi-
mental group
G
3
of the Ramos cell cultures. It is crucial to note that
human preparation factors involved in the RNA extraction from these
different cell lines necessarily introduce different degrees of variability.
For this reason, the following discussion simply demonstrates the
methodology for comparing replicate noise levels between differing
experiments. It is not intended to provide a definitive quantitative
comparison of the noise intrinsic to the MPSS and Affymetrix tech-
nologies. Such a comparison would require replicate measurements
using both technologies to be carried out on the same sample of target
RNA as well as a more precise calibration of the expression scales.
And, even in this case, the observed results would necessarily include
human factors present in the sample preparation steps of each of
the technologies. Even with these caveats, it is still interesting to have
a qualitative comparison of the repeatability of these quite different
gene expression technologies.
In figure 4.15 are replotted the values of s
(
q
)
plotted in figure 4.11b.
Also plotted are the values of s
(
q
)
from the Affymetrix data of
figure 4.3c, where q
,
dq, and s
(
q
)
have been rescaled so that the new plot
of s
(
q
0
)
versus q
is correctly expressed in units of log
10
(tpm). Note that
0
Search WWH ::
Custom Search