Sponsored by:
Location:
Leonard S. Klinck,
Room 460
University of British Columbia
Date:
Saturday, April 10
Contacts:
Camila P. Estevam de Souza
camilaestevambr@gmail.com
Joslin Goh
joslin_goh@sfu.ca
Welcome to the Spring 2010 Joint UBC/SFU Graduate Student Workshop in Statistics Webpage!
Many thanks to Pacific Institute of Mathematical Sciences for making this event possible.
Schedule
9.30am - 10.00am Coffee/Muffins
10.00am - 11.00am Paul Gustafon
11.00am - 11.30am Steven Wang
11.30am - 12.00pm Ryan Lekivetz
12.00pm - 12.30pm Stephanie Cheng
12.30pm - 1.30pm Lunch
1.30pm - 2.30pm Jiguo Cao
2.30pm - 3.00pm Darby Thompson
3.00pm - 3.15pm Break
3.15pm - 3.45pm Xuekui Zhang
3.45pm - 4.15pm Joslin Goh
5.30pm - Dinner
Abstracts
Speaker: Paul Gustafon
Title: The Publication Process in Statistics
Based on my experiences as an author, a referee, an Associate Editor,and an Editor, I will offer some thoughts about the publication process in our discipline. I will try to cover "the good, the bad, and the ugly" sides of the process.
Speaker: Steven Wang
Title: Joint inference for longitudinal and survival data with application to the AIDS studies Model
In many longitudinal studies, the individual characteristics associated with the repeated measures may be covariates for the time to an event of interest. Thus, it is desirable to model the time-to-event process and the longitudinal process jointly. Statistical analysis may be complicated with missing data or measurement errors in the time-dependent covariates. This article considers a nonlinear mixed-effects model for the longitudinal process and the Cox proportional hazards model for the time-to-event process. We provide a method for simultaneous likelihood inference for non-ignorable missing data and extend the method to time-dependent covariates. We adapt a Monte Carlo EM algorithm to estimate the model parameters. We compare the method with a naive two-step method and a bootstrap method with some interesting findings. An AIDS dataset is used as an illustration.
Speaker: Ryan Lekivetz
Title: Partially Clear Two Factor Interactions for Fractional Factorial Designs
In a 2^{m-p} design, a two-factor interaction (2fi) is said to be clear if it is not aliased with any main effect or any other 2fi. A clear 2fi can be estimated under the weak assumption that all interactions of three or more factors are negligible. However, the existence of a clear 2fi can be restrictive on the number of factors that can be used in a design. We examine the usefulness of "partially clear" 2fi's - those that can be estimated when certain 2fi's are also negligible, and report our findings for designs of 32 and 64 runs.
Speaker: Stephanie Cheng
Title: Osteoporotic hip fractures: incidence rates in BC and around the world.
Hip fractures causes more disability than any other type of osteoporotic fractures, which is of particular concern to industrialized countries with an aging population. In this talk, I will discuss two projects I was involved in during my co-op term at St. Paul's hospital, relating to hip fracture incidence rates. The first is an investigation into the hip fracture incidence rates in British Columbia, Canada over the past 15 years. We looked at crude hip fracture rates and assessed whether these rates were changing over time. The second project is a comprehensive literature review on global hip fracture incidence rates. Our aim, was to describe geographic and secular trends in incidence of hip fractures around the world.
Speaker: Jiguo Cao
Title:
It is not easy to obtain a master or PhD degree in statistics. Some of my experiences and mistakes will be shared with you. I will also talk about my life as a postdoc and junior faculty member.
Speaker: Darby Thompson
Title: Comparison of Models for Time-to-Event data subject to Multi-Phase hazards
This talk will focus on several methods for modeling time-to-event data subject to a multi-phase hazard. The term 'multi-phase' hazard is used when the hazard function for survival is not well modeled by parameteric functions, nor are the assumptions for standard semi-parametric (cox) models met and can arise when the hazard takes more complex shapes during different portions of the individual's survival experience. Examples of a hazard composed of multiple phases would be in a clinical trial when patients are subject to a traumatic intervention where the hazard function is composed of distinct phases risk of death due to the intervention itself being immediately very high (eg: immediately post-operative) until the survivors eventually receive benefit. Multi-phase hazards can also be found in clinical applications where covariates have a time-dependent effect such as a lag or erosion of the treatment-effect. If a treatment effect changes dramatically with time, the proportional hazards assumption necessary for a standard Cox model is violated. Several models which address this problem will be presented. The methods for fitting these models using common software tools (such as R or SAS) is described, and the fit and interpretation of the models is compared in two data sets. Models include fully and semi-parametric segmented (changepoint) models, spline models, and mixture models. Standard Weibull and Cox models are also fit for comparison.
Speaker: Xuekui Zhang
Title: Model-based method for analyzing next generation sequence data.
ChIP-seq, which combines chromatin immunoprecipitation with massively parallel short- read sequencing, can profile in vivo genome-wide transcription factor-DNA association with higher sensitivity, specificity and spatial resolution than ChIP-chip. While it presents new opportunities for research, ChIP-seq poses new challenges for statistical analysis that derive from the complexity of the biological systems characterized and the variability and biases in its digital sequence data. We propose a method called PICS (Probabilistic Inference for ChIP-seq) for extracting information from ChIP-seq aligned-read data in order to identify regions bound by transcription factors. PICS identifies enriched regions by modeling local concentrations of directional reads, and uses DNA fragment length prior information to discriminate closely adjacent binding events via a Bayesian hierarchical t-mixture model. Its per-event fragment length estimates also allow it to remove from analysis regions that have atypical lengths. PICS uses pre-calculated, whole-genome read mappability profiles and a truncated t-distribution to adjust binding event models for reads that are missing due to local genome repetitiveness. It estimates uncertainties in model parameters that can be used to define confidence regions on binding event locations and to filter estimates. Finally, PICS calculates a per-event enrichment score relative to a control sample, and can use a control sample to estimate a false discovery rate. We compared PICS to the alternative methods MACS, QuEST, and CisGenome, using published GABP and FOXA1 data sets from human cell lines, and found that PICS?? predicted binding sites were more consistent with computationally predicted binding motifs. If time permitted, I will briefly discuss how to modify the model proposed in PICS paper to analyze Histone modification data.
Speaker: Joslin Goh
Title: Optimal Fractional Factorial Split-plots Designs for Model Selection
Fractional factorial designs are used widely in screening experiments to identify significant effects. It is not always possible to perform the trials in a complete random order and hence, fractional factorial split-plot designs arise. In order to identify optimal fractional factorial split-plot designs in this setting, the Hellinger distance criterion (Bingham and Chipman (2007)) is adapted. The approach is Bayesian and directly incorporates common experimenter assumptions. By specifying prior distributions for the model space, the criterion for fractional factorial split-plot designs aims to discriminate between the most probable competing models. Techniques for evaluating the criterion and searching for optimal designs are proposed. The criterion is then illustrated through a few examples with further discussion on the choice of hyperparameters and flexibility of the criterion.
Directions
SFU to UBC: Take the Millenium Line Waterfront train to Commercial Skytrain station, then hop on to Bus 99B-line to get to UBC.
