Hello, my name is Trey Ideker,
I'm a professor in
the Department of Bioengineering at the University of California, San Diego.
The title of my talk today is protein networks and analysis of global gene expression.
Much of the recent excitement in network biology runs in parallel to and by analogy
to many of the successful developments in
protein sequence biology over the past 30 years.
Over here at left is shown what's by far
the most powerful paradigm in all of bioinformatics and perhaps, in all of genomics.
The idea is that if you have a critical mass of
DNA and protein sequence in a public database,
then you can do a very simple yet powerful things
like query that database with a short tidbit like
the nucleic acid sequence shown here to find all of the sequence similar matches
based on what's known about the function and structure of some of those matches.
Then you can infer that the structure and function
of your sequence query may, in fact, be the same.
What's shown over here now at right is
the analogy at the level of the protein interaction network.
What my group and several groups of other labs are trying to do,
is to develop an analogous set of
bioinformatic tools that would allow you to query not just the protein sequence database,
but databases of protein-protein and other kinds of interactions.
In this case, the goal is somewhat less clear,
but generally would be to query this database with
a global data set that's complimentary to the interaction data,
in order to find particular patterns of
interactions and regions of the interaction network that correspond to
different signaling and gene regulatory pathways of interest.
So what kind of interactions are,
in fact, being deposited in these large public databases?
The first kind of interaction I'll talk about is called a protein DNA interaction,
which means an interaction measured between
a protein transcription factor and a DNA promoter element.
Up here at the upper left is shown a network
taken from the work of Eric Davidson at Caltech,
who assembles large regulatory network models of sea urchin development.
One of the main methods that's just arisen in the past couple of years
for measuring protein DNA interactions rapidly at high-throughput,
has been developed by the labs of Richard Young and others,
and it's called a chromatin immunoprecipitation followed
by microarray chip hybridization, ChIP-chip for short.
The second kind of interaction that is being measured now
systematically are pairwise interactions among proteins,
the so-called protein-protein interactions.
These are then visualized in large networks of nodes and links,
where the links represent
physically measured binding event between the two proteins represented by nodes.
The two main methods now which are able to rapidly measure
protein-protein interactions are a genetic approach
known as the yeast two-hybrid system and a biochemical approach,
which involves co-immunoprecipitation, followed by mass spectrometry.
Finally, although I won't focus on them as much in my talk,
there are, of course, many other kinds of networks that are important to the cell.
One such network is, of course,
the biochemical reaction network or the metabolic network,
where the nodes are essentially metabolites and the links between
nodes would be metabolic reactions to convert one metabolite into the other.
We don't measure these at high-throughput these days.
Nevertheless, 50 years of biochemistry has resulted
in large network databases of metabolic reactions,
so we do have these at our fingertips in the public databases.
To give you an example of the biochemical techniques