Information Technology Reference
In-Depth Information
A great intellectual challenge in using new technologies is devising a way in which to
extract the full meaning of the data stored in large experimental gene expression
datasets. As “data mining” has been defined as “the exploration and analysis, by
automatic or semi-automatic methods, of large quantities of data in order to discover
meaningful patterns and rules”
[3]
. In brief, the goal is to convert data into
information and then information into knowledge
[4].
Freeing this knowledge is the
key to increase performance and success in the information age. Exactly this task is a
key challenge for bioinformatics. With the acceptance of a challenge of such a
complexity it is clear that also the solution to be provided by bioinformatics will be
very complex.
In this paper we want to address the complexity of this bioinformatic demand and
discuss some of the technical and intellectual issues involved in these processes,
describe some of the ways in which they are currently being addressed. We want to
introduce a bioinformatic approach combining experimental micro array data with
several methods such as database techniques, data mining, artificial intelligence,
statistics, modeling and computer graphics as shown in Fig 1. It is not done to use this
or that special method or to find the right one! The point is that by coupling several
proper methods in a well working bioinformatic software system it is expected to
achieve synergetic effects. Fig. 1 shows a proper combination of several methods of
different scientific domains to solve the complex task described above. Basic
components of such a complex bioinformatic working facility are a central data
storage and tools for analyzing and presenting data and results.
Analysis of
micro array
exression
data by
statistical
methods in
molecular
biology
Automatic
text mining
in online
journals
for
causal relations
between genes
and for protein
functions
Categorization
of typical
gene patterns
by
methods of
artificial
intelligence
Gene
product
function
prediction
by
causal
genetic
networks
and
statistical
methods
Visualization
of causal
genetic
networks
Experiments
Standardization
Normalization
Molecular
database
ŒExpression data
from experiments
ŒCausal relations
between genes
ŒClinical data
ŒGenetic networks:
Œ
Genes
Œ
Causal relations
Fig. 1.
Genome oriented bioinformatics at the Institute for Medical Informatics and Biometry,
University of Rostock
A molecular database stores micro array expression data from experiments in a
standardized and normalized form as well as deliverables from other system
