Button to scroll to the top of the page.



Student Spotlight: Tianjian Zhou


Meet Tianjian Zhou, a fourth-year PhD in Statistics student in the Department of Statistics and Data Sciences.

Tell us a little about yourself—educational background, previous work, experience, etc.

I’m from China. I received a BS in Statistics from University of Science and Technology of China, Special Class for the Gifted Young. After that, I continued my PhD study in Statistics at UT, which is a straightforward choice.

What attracted you to the PhD in Statistics program at UT?

I am always interested in exploring something that is totally new. I came to UT in 2013, when UT just started the PhD in Statistics program. This new program gave me the opportunity of becoming one of the first cohort of PhD students, which is exciting.

A more important reason is that UT SDS has prominent faculty. They serve as editors (or associate editors) of top journals and publish extensively. Their former PhD students usually have good placements.

Tell us about a project or piece of research you have worked on while attending UT.

I have been working on a project about Bayesian approach to tumor heterogeneity. During tumor growth, tumor cells acquire somatic mutations that allow it to gain advantages over time compared to normal cells. As such, tumor cell populations are oftentimes heterogeneous consisting of mixture of homogeneous cell subpopulations, a phenomenon called tumor heterogeneity. A subpopulation is characterized by unique genomic variants in its genome. We use next-generation sequencing data to infer such tumor heterogeneity. Nonparametric Bayesian inference approach enables us to flexibly model such data. I collaborate with my advisor and have spent two summers at NorthShore University HealthSystem working with other collaborators.

I am also working on other interesting projects, such as Bayesian approach to missing data in the presence of auxiliary variables, and Bayesian spatial modelling of khipu database.

You recently presented your research at a conference. Which conference did you present at? When?

I presented at iBRIGHT (Integrative Biostatistics Research for Imaging, Genomics, & High-throughput Technologies in Precision Medicine) 2015, which was held in November 2015 in Houston.

What is the title of your talk?

PairClone: A Bayesian Nonparametric Model for Reconstructing Tumor Subclones Based on Mutation Pairs

Do you have an analogy to help people understand your work?

Tumor cells contain different subpopulations. Imagine sand in different colors that is mixed together. The goal is to infer the number of colors and the proportion of sand having each color. We are only able to look at a grain of sand each time, and we might make random mistakes. Bayesian nonparametric approach flexibly models the number of colors and takes random noise into account.

Why is this research important?

My research can enrich our understanding significantly on tumor evolution and cancer development. Since tumor cells with different genomes probably response to different treatments, my research can also provide crucial information for development of precision medicine.

What happens next?

We are planning to incorporate other complexities in our model, including copy number variation, tumor purity and tumor phylogeny. Also, several classical assumptions about tumor heterogeneity are not necessarily true, such as normal cells are homogeneous, mutation occurs only once, etc. We are still investigating these questions.


Fun Facts

A talent you have always wanted: Piano

Favorite book: Norwegian Wood (Haruki Murakami)

Role model: Terence Tao, Eason Chan

If you weren’t in graduate school, what would you do? Work in some IT company.


SDS Tutorial Videos Reach 200,000 Views!
Student Spotlight - Sophia Yang Hooper

Related Posts