Peter J. Woolf on Bayesian Networks
Emerging Research Front Commentary, August 2010
|
Article: Bayesian analysis of signaling networks governing embryonic stem cell fate decisions
Authors: Woolf, PJ;Prudhomme, W;Daheron,
L;Daley, GQ;Lauffenburger, DA |
Peter J. Woolf talks with ScienceWatch.com and answers a few questions about this month's Emerging Research Front paper in the field of Computer Science.
Why do you think your paper is highly
cited?
I think the paper is highly cited because it was one of the first to provide a clear example of how Bayesian networks can be useful to biologists to interpret complicated experimental datasets.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
"We are only seeing the start of this now via social networking sites and distributed interactive media..."
The paper provides a new synthesis of knowledge by drawing from the fields of machine learning, stem cell biology, and bioinformatics.
Would you summarize the significance of your paper
in layman's terms?
This paper provides a clear example of how Bayesian networks can be used to visualize and interpret complicated cellular processes across length scales. Most uses of Bayesian networks in bioinformatics have focused on reconstructing large networks from homogenous data, such as gene expression data.
We extended this idea in our model to explicitly include the experimental perturbations, proteomic changes, and cellular outcomes. The resulting model provides a structured way to understand how each perturbation influences each step in the pathway.
How did you become involved in this research, and
how would you describe the particular challenges, setbacks, and
successes that you've encountered along the way?
I've always been interested in finding patterns in large datasets, and biological development has always fascinated me. As such, this paper represents a synthesis between these two fields for me.
Where do you see your research leading in the
future?
I see my research leading toward finding ways to better integrate human minds and machine learning in solving biological problems. Machines are superbly good at some things, but are woefully bad at some kinds of data integration. Rather than try to reproduce the human mind in silico, I see my work evolving into finding better ways to make researchers and their computers peers—each specialized in what they are good at.
Do you foresee any social or political
implications for your research?
Ha! Absolutely. We are only seeing the start of this now via social
networking sites and distributed interactive media. These tools only employ
the most basic machine learning approaches, but already they have changed
how we communicate both socially and scientifically.
Peter J. Woolf, Ph.D.
Chief Executive Officer
Foodwiki LLC
Web
KEYWORDS: BAYESIAN ANALYSIS, SIGNALING NETWORKS, EMBRYONIC STEM CELL FATE DECISIONS, ACTIVATED PROTEIN KINASE, SELF-RENEWAL, GENE EXPRESSION, P38 MAPK, DIFFERENTIATION, PHOSPHORYLATION, PROLIFERATION, PATHWAYS, STAT3, CONSTRAINTS, LEARNING ALGORITHM.