Genomic Data Explosion – The Challenge for Bioinformatics? - Advances in Data Mining - page 84

subset statistics for evaluating the significance of observed changes in the complete
dataset [7] .
Executable investigation levels of gene data analysis are genome, transcriptome,
proteome and metabolome. The last three levels can be distinguished from the first
one by context dependence. The entire complement of mRNA molecules, proteins or
metabolites in an organism, organ or tissue varies with it's physiological, pathological
or developmental condition. The analysis of transcriptome using mRNA expression
data is providing amounts of data about gene function, but it is an indirect approach
because mRNA are transmitters of genetic information, not functional cellular
entities. Comprehensive analysis of proteins and metabolites are more technical
demanding [8]. A summary for minimum information about a micro array experiment
(MIAME) now available in version 1.1 is developed by the object management group
(OMG). This standard gives the minimum information required to unambiguously
interpret and potentially verify array based gene expression monitoring experiments.
MIAME aims to define the core that is common to most experiments and is a set of
guidelines. A standard micro array data model and exchange format (MAGE) has
been recently developed by the OMG [9]. MIAME is used for standardizing our data
from micro array expression experiments.
To speed up the exploitation of human genome sequencing efforts, the European
Bioinformatics Institute (EBI) - an outstation of the European Molecular Bio-
logy Laboratory (EMBL) - is launching a publicly accessible repository for
DNA micro array-based gene expression data. ArrayExpress, a public database
( http://www.ebi.ac.uk/microarray/ArrayExpress/arrayexpress.html ) for micro array
based gene expression data, supports all the requirements specified by the MIAME
and aims to store MIAME compliant data. This database will allow to cross-validate
data obtained by different technologies.
2.3 Data Analysis
A variety of techniques has established to monitor the abundance for all of an
organism's genes [10], [11], [12]. Some of them should be considered:
Average Difference (Avg Diff) - a Parameter Used for Analyzing Expression
Data. As we use the technology of Affymetrix™ (glass slides, fluorescent-labelled
cDNA) we want to introduce one of the resulting analysis parameters of data
processing which includes information about fluorescence intensity - the average
difference. This parameter can be interpreted as a gene expression value at mRNA
level. That is the reason why this parameter is often used for data analysis of
expression experiments. But if we look at this parameter in detail we will find, that
there are some points to be noticed. Observing the statistical distribution of this
parameter we will note that there are no values in the range of (-1, +1). The sector
around null is not specified by the system. Furthermore we found an unsymmetrical
distribution comparing two nearly identical expression experiments. The outcome of
this is the question whether this is a systematic error based on two independent
analysing algorithms for the positive and negative range or maybe the data processing
method is extremely sensitive to minimal influences. At least these points are to be
kept in mind when using the avg diff - and only this parameter - for getting analysing
results of gene expression data. A practicable example for using the avg diff by