09:30-9:50: Coffee Break
9:50-10:05: Open Remarks & introduction

Session 1: Presentations

Chair: Geneviève Lefebvre (UQAM)

Talk 1: Juli Atherton (McGill)


Using SELEX Data to Model the Affinity of DNA Sequences to the Transcription Factor Bicoid

In genomic research, determining locations on the genome to which a transcription factor binds with medium to high affinity might help identify possible transcription factor binding sites. Hence, interest lies in developing models that predict the affinity of a transcription factor based on nucleotide sequence. Often, data from in-vitro experiments are used when building such models. One such in-vitro experiment is a systematic evolution of ligands by exponential enrichment (SELEX) experiment. In a SELEX experiment one begins with a large random pool of DNA sequences in equilibrium with a transcription factor. Sequences that had bonded to the transcription factor are separated from the solution, amplified by polymerase chain reaction (PCR) and entered into the next round of SELEX. This process continues for as many rounds as the experimenter desires. Thus far SELEX has been very good at suggesting consensus sequences but making further inference from them has been difficult. In this talk, I will begin with a simple biochemical explanation of SELEX. I will then discuss our analysis of the SELEX data for the transcription factor Bicoid. This is a joint effort with Mark Bigginís lab (Lawrence Berkeley National Laboratory) and Peter Bickelís group (Department of Statistics, UC Berkeley). The data are part the Berkeley Drosophila Transcription Network Project.

Talk 2: Lilia Leticia Ramirez Ramirez (Waterloo)


Dynamics of Infectious Diseases in Networks

The epidemic models studied here relax the common epidemic assumptions that susceptible individuals are equally likely to acquire the disease. A structure for the social interactions that is important for the disease transmission is incorporated in the population. This contact configuration can be very heterogeneous (as in small worlds) and it is modeled as a random graph whose edges describe the kind of contacts between individuals that can result in infection.

In contrast with more common epidemic models, here the latent and infectious period can have distributions other than exponential and the rate of contacts can also be a random variable.

This work extends the epidemic models suggested by Newman (2002) in two lines. The first, studies the hierarchical networks that model the interaction within and between sub-populations. The second direction examines the outbreak evolution in discrete time.

The results here obtained can be directly applied to study the dynamics of other kind of "agents" such as information and ideas. For example, the dynamics can involve the spread of computer viruses, rumors, and personal positions regarding a factor idea.

10:45-11:15: Coffee Break (Getting to know each other)

Talk 3: Lajmi Lakhal-Chaieb (Laval)

Abstract: Nonparametric Estimation of Kendall's Tau for Serial Censored Gap Times

In this talk, we present a nonparametric estimation of Kendall's tau for serial gap times. Such setting typically involves a dependent censoring. We present an IPCW estimator to recover dependent censoring. The proposal is illustrated by simulations and with a real data set on colon cancer.

Joint work with Richard J Cook (Waterloo) and Xihong Lin (Harvard).

Talk 4: Taoufik Bouezmarni (Montréal)

Abstract: Nonparametric Beta Kernel Estimator for Long Memory Time Series

The nonparametric smoothing of the periodogram of a long memory time series is considered. In contrast to most previous methods, we propose a Beta kernel estimator that is suited to the shape of the spectral density of long memory processes. In particular, the estimator is automatically adapted to the boundness or unboundness regions of the spectral density. Consistency of the estimator is established, as well as empirical illustrations on simulations and real data.

Talk 5: Azadeh Moghtaderi (Queen's)
Abstract: A Novel Estimator of the Wold-Cramèr Evolutionary Spectrum

A nonstationary stochastic process is one whose statistical properties are time-dependent. That is, its mean, variance, covariance function, and higher-order moments may change as time evolves. The Wold-Cramèr evolutionary spectrum (WCES), introduced by Mélard (1975), is a time-dependent analogue of the spectrum of a stationary stochastic process which is valid for all nonstationary stochastic processes. Existing estimators of the WCES suffer mainly from two problems: (i) Bias in the boundary regions of the time-frequency plane, and (ii) a trade-off between time and frequency resolution. We propose a novel estimator of the WCES which mitigates problem (i) by extrapolating the WCES in time using an estimate of its time derivative. In the case that the nonstationary stochastic process under study is uniformly modulated (UM), we show that our estimator can be modified to produce an estimator with ìgoodî time and frequency resolutions. This mitigates problem (ii) for UM nonstationary stochastic processes. We demonstrate the performance of our estimators on simulated nonstationary stochastic processes whose Wold-CramÈr evolutionary spectra are known analytically.

12:15-13:15 Lunch Break

Session 2: Round Table Discussions

13:15-14:15 Identifying Important Issues Relevant to New Investigators

Chair: Yulia Gel (Waterloo)

14:15-15:15 Group Discussions

Chair: Juli Atherton (McGill)

15:15 -15:35 Coffee Break

15:35-16:35 Final Discussions/Closing Remarks
Chair: Jason Nielsen (Carleton)