Skip to main content

Fall 2020 Virtual Seminar Series

Department of Statistics 2020-2021 Seminar Series (joint with Biostatistics) - Fall 2020

The Fall 2020 Seminar Series (joint with Biostatistics) talks will be offered virtually using Zoom instead of in person. Registration will be required to receive the zoom link for the event. Please see the registration link associated with each talk to sign up. Links are specific to individual talks. Please email Kisa Kowal at k-kowal@northwestern.edu if you have questions. 

Seminar Series talks are free and open to faculty, graduate students, and advanced undergraduate students.

D-optimal Designs for Multinomial Logistic Models

Wednesday, September 30, 2020

Time: 11:00 a.m. central time - registration required

Speaker: Jie Yang, Professor, Department of Mathematics, Statistics, and Computer Science at University of Illinois at Chicago

Abstract: Design of experiment with categorical responses is becoming increasingly popular in a rich variety of scientific disciplines. When the response is binary, generalized linear models have been widely used. For optimal designs with generalized linear models, the minimum number of distinct experimental settings required by a nondegenerate Fisher information matrix is equal to the number of parameters. It is also known that the experimental units should be uniformly allocated when a minimally supported design is adopted. When the response has three or more categories, the models used in the literature should rather be treated as a special class of the multivariate generalized linear models, known as multinomial logistic models. We show that, unlike the designs for binary responses, a feasible design for a multinomial logistic model may contain less experimental settings than parameters, which is of practical significance. We also conclude that even for a minimally supported design, a uniform allocation, which is typically used in practice, is not optimal in general for a multinomial logistic model. We develop efficient algorithms for searching D-optimal designs. Using examples based on real experiments, we show that the efficiency of an experiment can be significantly improved if our designs are adopted.

Zoom Registration link

 

Trends in Hiring Discrimination against Racial and Ethnic Minorities in Western Countries

Wednesday, October 14, 2019

Time: 11:00 a.m. central time - registration required

Speaker: Lincoln Quillian, Professor of Sociology and Faculty Fellow, Institute for Policy Research at Northwestern University

Abstract: In this presentation, I examine trends in hiring discrimination against racial and ethnic minorities over the last 20 to 30 years in countries in North America and Europe. The analysis uses estimates of discrimination from about 120 field experiments of hiring discrimination. In these experiments falsified applications from members of different racial and ethnic groups are made for actual job listings. The racial disparity in positive responses is used as a measure of discrimination. I find few changes over time in hiring discrimination in both pooled samples of countries and in results for several individual countries and target groups. There are a few exceptions, most notably that discrimination in France may have declined (but remains high). I analyze factors predicted by theory to affect trends in hiring discrimination. Employment discrimination appears to be driven by relatively stable biases against non-white applicants that are only weakly linked to political attitudes expressed on surveys. The presentation will also discuss the strengths and weaknesses of using a time series of field experiments to understand trends.

Zoom Registration Link

 

Bayesian Modeling of Metagenomic Sequencing Data for Differential Abundance Analysis

Wednesday, October 28, 2020

Time: 11:00 a.m. central time - registration required

Speaker: Qiwei Li, Assistant Professor of Statistics, Department of Mathematical Science, The University of Texas at Dallas

Abstract: Advances in next-generation sequencing technology have enabled the high-throughput profiling of metagenomes and accelerated the study of the microbiome. One of the most essential questions that can help us decipher the relationship between the microbiome and disease is how to identify differentially abundant taxonomic features across different populations. Metagenomics sequencing data are usually summarized into a high-dimensional count table, which suffers from sample heterogeneity, unknown mean-variance structure, and excess zeros. These characteristics often hamper downstream analysis and thus require specialized analytical models. In this paper, we propose a Bayesian bi-level framework to identify a set of differentially abundant taxa, which could potentially serve as microbial biomarkers for diagnosing diseases. The bottom-level is a multivariate count generative model that links the observed counts in each sample to their latent normalized abundances. For the choice of a zero-inflated negative binomial model as the bottom level, we use the Dirichlet process as a flexible nonparametric mixing distribution to model all latent factors that account for sample heterogeneity. The top-level is a Gaussian mixture model with a feature selection scheme for identifying those taxa whose normalized abundances are discriminatory between different phenotype groups. The model further employs Markov random field priors to incorporate taxonomic tree information to identify microbial biomarkers at different taxonomic ranks. A colorectal cancer case study demonstrates that a resulting diagnostic model trained by the microbial signatures identified by our model in a cohort can significantly improve the current predictive performance in another independent cohort. In summary, this statistical methodology provides a new tool for facilitating advanced microbiome studies and elucidating disease etiology.

 Zoom Registration Link

Feature Selection with Survival Outcome Data

Wednesday, November 4, 2020

Co-hosted by Women in Statistics (WIST)

Time: 11:00 a.m. central time - registration required

Speaker: Hyokyoung (Grace) Hong, PhD, Associate Professor, Department of Statistics and Probability, Michigan State University

Abstract: Detecting biomarkers that are relevant to patients' survival outcomes is crucial for precision medicine. Dimension reduction is key in the process. Although regularization methods have been used for dimension reduction, they cannot handle a large number of candidate biomarkers generated by modern bio-techniques. Variable screening, which has been widely used for handling exceedingly large numbers of variables, is however much underdeveloped for censored outcome data. This talk introduces a series of new feature screening procedures that I have recently developed for survival data with ultrahigh dimensional covariates. These methods include conditional screening, integrated powered density screening, Lq-norm learning, and forward regression with partial likelihood. I will discuss the intuition behind and demonstrate their utilities through real data analyses.

 Zoom Registration Link

 

Multiscale Inference and Modeling of Cell Fate via Single-cell Data

Wednesday, November 11, 2020

Co-hosted with the NSF-Simons Center for Quantitative Biology

Time: 11:00 a.m. central time - registration required

Speaker: Qing Nie, Department of Mathematics, Department of Developmental and Cell Biology, NSF-Simons Center for Multiscale Cell Fate Research, University of California, Irvine

Abstract: Cells make fate decisions in response to dynamic environmental and pathological stimuli as well as cell-to-cell communications.  Recent technological breakthroughs have enabled to gather data in previously unthinkable quantities at single cell level, starting to suggest that cell fate decision is much more complex, dynamic, and stochastic than previously recognized. Multiscale interactions, sometimes through cell-cell communications, play a critical role in cell decision-making. Dissecting cellular dynamics emerging from molecular and genomic scale in single-cell demands novel computational tools and multiscale models. In this talk, through multiple biological examples we will present our recent effort to use single-cell RNA-seq data and spatial imaging data to uncover new insights in development, regeneration, and cancers. We will also present several new computational tools and mathematical modeling methods that are required to study the complex and dynamic cell fate process through the lens of single cells.

Zoom Registration Link

Back to top