Information Technology Reference
In-Depth Information
integrating into the expression data analysis is given by
[13].
Gene expression analysis
of oligo nucleotide micro arrays to determine gene expression profiles of the inflamed
spinal cords of experimental autoimmune encephalomyelitis (EAE) mice at the onset
and at the peak of the disease are described. Of the approximately 11 000 genes
studied, 213 were regulated differentially and 100 showed consistent differential
regulation throughout the disease. These results are obtained using among others the
avg diff and several clustering methods in the data processing phase.
Graphical Plot - often Called Clustering.
A feature of gene expression is the
tendency of expression data to organize genes into functional categories. It is not very
surprising that genes that are expressed together are sharing common functions. So we
can cluster genes if they are expressed with the same expression profile under same
conditions. But what is that - clustering genes? And how to do that right? We find the
word clustering nearly in every publication about micro array data analysis. Often this
word is used in a very common context, e.g. grouping genes by their expression level
or by any other parameter they are labelled with. View the expression profile under
specified experimental conditions or under time. Independent of whether the data sets
originate from drug responses, molecular anatomy studies or disease models, any
analysis starts with a grouping of expression patterns according to their similarity and
existing annotations. So the next challenge in expression analysis lies in comparing
diverse data sets for which it would not make sense to analyse per cluster analysis the
data together
a priori
. Graphical tools can help here to illustrate more-complex
relationships. Often is meant a kind of grouping (clustering) with a graphical
approach. The result will be a graphical presentation of similar profiles in time, space
or under special experimental conditions like concentration gradient.
Cluster Analysis.
As described above we will find several approaches of grouping
(clustering) data, but it is a difference whether making a clustering which means a
graphical grouping or a clustering which means cluster analysis, actually coming from
the field of multivariate statistics. The term “clustering” is applied in both contexts.
But the idea in the latter case is to process a cluster analysis. This is an exact defined
statistical domain with several methods to be practised. Cluster analysis is a set of
methods for constructing a (hopefully) sensible and informative classification of an
initially unclassified set of data, using the variable values observed on each individual
[14]
. Given a sample of genes with an expression value we are looking for a
characterization of potential clusters of genes and a state which gene is to be assigned
to which cluster. We will get a result of statistical analysing algorithm. This could be
a cluster schedule in form of a table or list of all genes according each to a cluster
number and an according coefficient which gives a value for the distance to the next
(nearest) cluster. In most cases there is also given a graphical plot, a dendrogram in
case of hierarchical cluster analysis e.g. A very efficient software which performs
both variants of “clustering” - various graphical approaches and cluster analysis tools
too - is for example the GeneSpring software by Silicon Genetics available at
http://www.sigenetics.com/cgi/SiG.cgi/index.smf
.
Another system for processing cluster analysis of genome-wide expression data from
DNA micro array hybridization is described in
[4]
. This system uses standard
statistical algorithms to arrange genes according to similarity in pattern of gene
expression. Although various clustering methods can usefully organize tables of gene
expression measurements, the resulting ordered but still massive collection of
numbers remains difficult to assimilate. Therefore it is useful to combine clustering
