326
6. ANALYSIS OF DATA
After processing of the digital image signal, spot
objects can be recognized (genes, cDNAs, ESTs
or oligonucleotides). Each object owns more
groups of the same quantitative features, one
group for each cell population in the
experiment. For example the expression levels
measured for more cell populations in single
color mode (with more membranes and by using
internal standards) or the three color
components in dual-color mode detection (Chen
et al., 1998). Statistical analysis techniques
(estimation of data correlation between different
experiments, estimation of data scattering,
estimation of regression, etc.) and cluster
analysis (to recognize possible outliers in order
to study the differential expression among genes)
can be applied to these objects.
(Chen et al., 1998) studied two lung
adenocarcinoma cell lines in dual-color mode;
the two populations corresponds to two
different colors (red and blue) developed by
using two different color-forming enzymes.
After signal processing, a single color value was
associated to each target on the nylon
membrane; this color value was decomposed
into the three subtractive primary colors (Cyan
C, Magenta M and Yellow Y). Each target was
depicted as a point in the CMY space; the whole
panel of target was represented with a 3D-scatter
plot in the CMY space. In these conditions the
magenta-colored points represented genes
equally expressed in the two studied
populations. The red-colored and blue-colored
points represented instead a differential
expression of the corresponding target in the
two populations. It was possible to calculate a
regression line along which the points arranged
themselves; therefore it was possible to
determine outliers (showing the differential
expression of the target) compared to the
regression line. According to the authors, the
threshold to distinguish outliers was 99%
prediction interval in statistics of detection in
dual-color mode for two identical amounts of
the same cellular population labeled with the
two different enzymes.
7. CONCLUSIONS
Future technological progresses in the field of
laboratory experimentation, of image acquisition
and image processing, will make this approach
automated and faster.
Researching efforts will tend to enhance the
numerousness of the target panel and of the
cellular populations for a single image.
In this paper the general problems of this sort of
method, from the image acquisition to the
statistical analysis of final results has been
described.
A prototype of gene expression image simulator
has been developed to better understand all the
potential components affecting the gene
expression image processing techniques.
ACKNOWLEDGEMENTS
Illustrations created by John Hatton (Consiglio
Nazionale delle Ricerche - ITBA), and special
thanks for his revising of this manuscript.
REFERENCES
Bernard, K., Auphan, N., Granjeaud, S.,
Victorero, G., Schmitt-Verhulst, A. M., Jordan,
B. R., and Nguyen, C. (1996). Multiplex
messenger assay: simultaneous, quantitative
measurement of expression for many genes in
the context of T cell activation. In: Nucleic Acid
Res. 24: 1453-1443.
Chen, J. J. W., Wu, R., Yang, P-C., Huang, J-Y.,
Sher, Y-P., Han, M-H., Kao, W-C., Lee, P-J.,
Chiu, T. F., Chang, F., Chu, Y-W., Wu, C-W,
and Peck, K. (1998). Profiling expression
patterns and isolating differentially expressed
genes by cDNA microarray system with
colorimetry detection. In: Genomics 51: 313-324.
DeRisi, J. L., Penland, L., Brown, P. O., Bittner,
M. L., Meltzer, P. S., Ray, M., Chen, Y., Su, Y.
A., and Trent, J. M. (1996). Use of a cDNA
microarray to analyze gene expression patterns
in human cancer. In: Nat. Genet. 14: 457-460.
DeRisi, J. L., Iyer, V. R., and Brown, P. O.
(1997). Exploring the metabolic and genetic
control of gene expression on a genomic scale.
In: Science 278: 680-686.
Lockhart, D. J., Dong, H., Byrne, M. C.,
Follettie, M. T., Gallo, M. V., Chee, M. S.,
Mittmann, M., Wang, C., Kobayashi, M.,
Horton, H., and Brown, E. L. (1996).
