Biology Reference
In-Depth Information
Spot Position
Fluorescence
Ratio
Log
Red/
Green
Row
Column
Gene ''Name''
Red
Green
(Base 2)
1
2
A
1,000
100
10
3.32
1
5
B
120
1,200
0.1
-3.32
2
1
C
100
1,000
0.1
-3.32
2
3
D
1,500
750
2
1.00
2
4
E
150
1,500
0.1
-3.32
2
5
F
1,800
1,200
1.5
0.58
3
2
G
600
1,200
0.5
-1.00
3
3
H
800
80
10
3.32
4
1
I
1,500
1,500
1
0.00
4
3
J
50
500
0.1
-3.32
4
5
K
1,800
360
5
2.32
5
1
L
900
90
10
3.32
5
2
M
180
900
0.2
-2.32
5
4
N
1,000
1,500
0.67
-0.58
TABLE 12-2.
Synthetic data for Figure 12-7(A), with logarithmic transformation of red/green ratios.
E
XERCISE
12-1
Compare the ratio and logarithmic data from Table 12-2. Characterize
each as a decrease or an increase in expression. Why is using the log base
2 a better situation than using the ratios?
Note that in Tables 12-1 and 12-2 we did not list the features from
Figure 12-7(A) that had shown no fluorescence at all. This procedure is
called filtering. In a real microarray experiment, we might also want to
remove all of the features that had very low values. This type of filtering
process would reduce the size of the data set (thus increasing processing
speed) and remove the lowest-quality data because any spots with
intensity near the background level are likely to be measured with
questionable accuracy. Genes with missing values in replicate
measurements may also be filtered. Finally, depending on the
experiment, it may be important to only consider genes that changed
expression by a given amount, such as a factor of two. As with any
preprocessing of data, filtering may result in loss of information. When
carefully used, however, it increases processing speed and accuracy
without a significant risk of eliminating any important genes.
Through image processing, background correction, and filtering, the
information from the microarray that was initially stored as an image is
converted to a table of values for each gene present on the microarray.
