Biomedical Engineering Reference
In-Depth Information
of its being false positive is likely low. For example, the gene chip has two
probe sets with the same gene name, ANGPT2. Both of them (i.e., 1951 at
and 37461 at) were selected here as down-regulated genes with 1:581:59
fold changes. This information can be used to indicate the likelihood of true
dierential expression. If investigators want to know more detailed infor-
mation of the gene (e.g., DKK1), a click of this gene's probe set ID leads to
Figure 6.3, which lists most important gene properties, such as alias, locus
location, summary, GO annotations, pathway, and reference into function.
A scatterplot was given to display data distribution (e.g., gene expressions
were well separated for the gene, DKK1). A link to NCBI's Entrez Gene
database is given by clicking the Entrez gene ID if investigators want to
know most of the detailed information.
To check the results of gene function classication, we can click the
cells of the table of \Number of Annotations" in Figure 6.1. For example,
Figure 6.4 illustrates the results of pathway classication for the down-
regulated and up-regulated genes. A higher frequency may indicate higher
likelihood of the corresponding pathway involved in the experiment. A click
of a frequency in the cell of the table will generate a table of listed genes. For
example, there are 3 selected genes associated with cell cycle in Figure 6.5,
which displays gene expression data, gene name, and gene description to
check fold change.
3.5. RT-PCR Validation
Using the integrated bioinformatics tool, the results of gene function classi-
cation led us to identify 6 genes strongly associated with cell proliferation
in Table 1. Five of the 6 genes were veried by quantitative RT-PCR. Four
of the ve genes were conrmed to be dierentially expressed in the 23k
PRL group. Only one gene, TB1, was misclassied as an up-regulated gene.
In the 16k PRL group, only one gene (Asparagine) was a false negative.
4. Discussion
In summary, microarray data analysis is a complicated process. It requires
multiple steps in order to yield more comprehensive results. First, we have
to check data quality. We describe the 2D image plot to ensure the high
quality of microarray array data for analysis. Once we have good quality of
data, we perform gene selection to identify dierentially expressed genes.
We present a probe rank approach to analyze probe level data which has the
advantage over the gene level data analysis, such as detection of alternative
Search WWH ::
Custom Search