David Bickel

David Bickel

Associate Professor
Department of Biochemistry, Microbiology and Immunology
Faculty of Medicine, University of Ottawa

Roger Guindon Hall, Room 4510F
451 Smyth Road, Ottawa, ON K1H 8M5

Tel: 613 562-5800 ext. 8670 (office)
Tel: 613 562-5800 ext. 8673 (Lab)
Email: d bickel [at] uottawa.ca

personal website
current lab members

We currently have the following opportunities see jobs section.

Gene Network Reconstruction:
Software for "Gene network reconstruction from transcriptional dynamics under kinetic model uncertainty"

David Bickel
Associate Professor

Degrees:

B.Sc. Baylor University 1993
M.A. University of North Texas 1994
Ph.D. University of North Texas 1997

Research Projects

My statistical bioinformatics research interests span gene expression data analysis, molecular network reconstruction, model validation methodology, Bayesian and empirical Bayes inference, machine learning algorithms, and Monte Carlo simulation.

• Bayesian integration of data from different sources. Genomic, proteomic, and metabolomic data pose formidable problems requiring statistical inference, especially when different types of data are considered jointly. The conventional hypothesis tests commonly used in data integration have well-known inconsistencies that often lead to unsupported conclusions. The Bayesian paradigm instead enforces self-consistency by stipulating that degrees of confidence placed in hypotheses and parameter intervals follow the laws of probability. Bayesian probability theory propagates uncertainty involved in each source of information to give the degree to which a given hypothesis is supported by various types of data and biological judgment.

Bayesian data integration poses conceptual challenges in modelling degrees of confidence in scientific hypotheses. Additionally, the volume and heterogeneity of data call for creative computational and mathematical strategies to approximate probabilities in a reasonable amount of time. Another fertile research area important for data integration is the interface between Bayesian statistics and methods recently developed by the machine learning community.

• Bayesian inference in biological systems. Probabilistic data integration is beginning to furnish the tools needed to honestly quantify the credibility of reconstructed biological networks. Although systems biology is typically seen in terms of cell biology and computer science, its complexity poses challenging problems in inductive reasoning that require innovations in stochastic process parameter inference and in applied statistics. In particular, the Bayesian framework can connect scientific knowledge to the mathematical modeling to be implemented in software, thereby facilitating appropriate consideration of the high degrees of uncertainty inherent in the study of biological systems. This steers a practical course between putting undue confidence in an optimized network and timidly controlling conventional error rates.

Selected Publications:

D. R. Bickel, Z. Montazeri, P.-C. Hsieh, M. Beatty, S. J. Lawit, and N. J. Bate, "Gene network reconstruction from transcriptional dynamics under kinetic model uncertainty: A case for the second derivative," Bioinformatics 25, 772-779 (2009). Full paper
D. R. Bickel, "Probabilities of spurious connections in gene networks: Application to expression time series," Bioinformatics 21, 1121-1128 (2005).
D. R. Bickel, "Degrees of differential gene expression: Detecting biologically significant expression differences and estimating their magnitudes," Bioinformatics 20, 682-688 (2004).
D. R. Bickel, "Robust cluster analysis of microarray gene expression data with the number of clusters determined biologically," Bioinformatics 19, 818-824 (2003).
D. R. Bickel, "Error-rate and decision-theoretic methods of multiple testing: Which genes have high objective probabilities of differential expression?" Statistical Applications in Genetics and Molecular Biology 3(1) 8, http://www.bepress.com/sagmb/vol3/iss1/art8 (2004).