Drug target prediction: Finding biological needles in a haystack of
networks
</br>
Eric Kolaczyk, Boston University
</br>
</br>
Understanding the mechanism of action of putative drug compounds and
locating potential genetic drug targets is a major focus in biomedicine.
Compendia of mRNA microarray expression data, generated under various
experimental conditions, and fairly readily available, can be a key source
of information in creating computational-based approaches to this problem.
However, there are a host of challenges to be faced in trying to do so.
In this talk, I will present an overview of work being done by our group
in this area on one major aspect of the overall problem -- the difficulty
of assessing in expression data the response of gene targets to
experimental perturbations against the `background' associated with
`typical' cell activity.  The approach we are developing is multifaceted,
involving methods for compendia design and choice of experimental
protocol, network data integration, and statistical modeling of sparse
networks. I will focus particularly on our contributions to this last
topic and describe a process of `network filtering' that shows promise for
extracting sparse signals of potential gene targets from a `background' of
gene regulatory interactions.
</br>
</br>
<a href="http://www.stat.columbia.edu/~tyler/seminar" title="Schedule">Back to the schedule</a>