Sample A: Cover Page of Thesis, Project, or Dissertation Proposal

In summary, each probe demonstrates its own binding affinities, and, since on Affymetrix arrays
there is no replication of individual probes, variation per hybridization reaction cannot be
assessed. Replication is represented solely in the probeset’s (transcript level) distribution of signal
intensities, and this distribution of the probe intensities then is represented with any of a number
of statistical estimators [52]. The aggregation of these probesets into a scalar value representing
the transcript concentration can be performed by mean analysis, trimmed mean analysis, median
analysis, etc., and there exists considerable debate about the appropriate methodology and
whether the mismatch probe information should be used, and how [45, 46, 48]. However, one
must acknowledge the dubious nature of a simple interpretation of these ‘measurements’ for
transcript levels when they are only pixel summarizations of fluorescence of sequence subsets of
the transcript, where no replication of spots occurs.
Data Analysis and the Impact of the Data Cleansing Methods on
Interpretation
The end result of a half-million individual Microarray reactions is the fluorescent intensity of a
grid of spots, which the scanner (or imager) records as a Microarray image. Software and
algorithms continue to be developed to accommodate adjustments for local and global
background spotting intensities and to extract intensities and merge them into scalar gene values
[52]. This interpretation of the signal represents a semi-quantitative measurement of the
transcript region’s concentration [7]. The investigator is then left to make decisions as to how to
aggregate individual probes into probesets, including whether to combine probesets that measure
the same transcript and/or the same gene [9]. An unexplored aspect of these experiments is
18