Biostatistics/Epidemiology Training Grants in AIDS
 


Fellows






FELLOWS

Current Fellows (2012-2013)

Kevin Galinsky, Daniel Schlauch, Sixing Chen, David Valken-LeDuc, and Ryan Sun are first year students in the Biostatistics doctoral program. They are currently focused on coursework, and are currently taking core biostatistics and epidemiology courses. They are currently working on summer projects:

  • Kevin will be working with Dr. Alkes Price on an estimator for PCA eigenvectors. Dr. Price is the author of EIGENSTRAT1 , an algorithm and software package that uses the first few eigenvectors from PCA to correct for population stratification in GWAS. The issue with PCA is that it is time consuming to run, with the most time consuming step being the computation of the covariance matrix of the genomes of all the subjects in the study.
  • Daniel plans to explore the role of measured clinical covariates in a newly published breast cancer dataset. Among other techniques, this will involve building regression models for the prediction of molecular outcomes such as copy number variation, methylation, and expression of RNA and adjusting for instances of confounding. He will also be using various novel classification techniques to attempt to find new molecular subtypes and gene signature predictors in breast cancer.
  • Sixing will be working on a research project on sequencing error with Dr. Xihong Lin. Sixing will correct the type I error in testing in the presence of sequencing error, as well as to examining the power.
  • David's research will focus on modeling the exposure-response relationship between aircraft noise and cardiovascular disease. He will be analyzing highly unbalanced cluster data provided by the Federal Aviation Administration which will be combined with Medicare data. In contrast to previous work in this area which focused on a linear relationship between exposure-response, he will be considering a non-linear relationship. During this process, David will be learning about Bayesian hierarchical models, change point models, random effect models and the use of splines. Using these methods, he plans to build a model or series of models that can detect and estimate the non-linear association of aircraft noise attributed to individual airports and the risk of cardiovascular disease.
  • Ryan will be working with Dr. Xihong Lin on detecting gene-environment interactions in genome wide association studies (GWAS). While the statistical community has made much progress in identifying genetic risk factors from GWAS (for breast cancer, lung cancer, etc.), there have been very few replicable results identifying possible gene-environment interaction risks (i.e. genes which enhance susceptibility to smoking). One of the problems in testing for GxE interactions is that the correct Type I and Type II error rates under hypothesis testing are not preserved when we test in the regression framework. This project hypothesizes that this is because of linkage disequilibrium (LD) between genes which are located close to each other. It is their hope that they can characterize the extent to which p-values are transformed due to the LD and offer a correction for future GxE tests.

Adam Sullivan is a third year student (dissertation advisor: Tyler Vanderweele) in the Biostatistics doctoral program. His research this past year has focused on analyzing data for three different papers dealing with multiple aspects of Religious and Spiritual Care given to end of life cancer patients. This research examined the perceptions of religious and spiritual care from the vantage of patients, nurses and physicians. One specifically centered on the impact of a dedicated service on palliative care. Another focused on how radiation oncologists evaluate and incorporate life expectancy estimates into the treatment of palliative cancer patients. Lastly, Adam is analyzing data concerning subjects who have had a metastatic cancer tumor removed. This project then links how much chemotherapy was given to make sure the tumor did not grow back and the time until the tumor returns. The project will also incorporate the amount of chemotherapy given as well as the extent of the area around the tumor that was treated.

Ian Barnett is a third year student (dissertation advisor: Xihong Lin) in the Biostatistics doctoral program. He is currently working on bringing the higher criticism test, original designed for high dimensional data to detect sparse signals, to the SNP-set genetic association testing realm. This is useful when detecting when there are only a few mutations related to disease within a gene. He is applying this new method to lung cancer GWAS and breast cancer GWAS data sets in order to identity the top associated genes.

Ian's previous project designed an SNP-set association test for studies that use extreme phenotype sampling when the outcome of interest is normally distributed. For this study, Ian gave a poster at the 2012 ASHG annual conference and was selected as a finalist for Harvard Horizons in March 2013.

Christina McIntosh is a fourth year student (dissertation advisor: Giovanni Parmigiani) in the Biostatistics doctoral program. She is currently working on a project that compares two methods of analyzing pedigrees for an association between a disease and gene. This project assesses the empirical power of the Family Based Association Test (FBAT) and the Pedigree Based Association Test (PBAT). Also, the empirical power of the ascertainment conditions of the pedigrees are compared. Additionally, she is working on expanding a survival model that involves twins to a Bayesian non-parametric framework. A new project that involves quantifying cancer "immunity" from a statistical framework.

Danielle Braun is a fifth year student (dissertation advisor: Giovanni Parmigiani) in the Biostatistics doctoral program. Currently, she is working on a project to develop methods for handling misreported family history in Mendelian risk prediction models for cancer. This project introduces an approach that allows adjustment for misreporting of family history by modeling the measurement error process in the survival context, and using this to weigh the risk prediction estimates. The team proposes different models for the measurement error process using data from UCI. The goal is to extend BRCAPRO, a Mendelian risk prediction model for breast and ovarian cancer, to handle misreported family history. They will extend BRCAPRO using the methods they develop. They also will illustrate the results using specific cases, as well as validate using CFR data.

Norman Huang is a fifth year student (dissertation advisor: Cheng Li) in the Biostatistics doctoral program. He has completed all his courses and has just defended his dissertation. Norman's dissertation primarily concerns how gene expression and pathway information can be used to predict patient response in MM patients. Predicting patient response has a significant amount of clinical value since better response rates naturally correlates with better survival. However, while the utility to predict response is clear, the main problem is that it is difficult to do so. Much of the problem arises from the fact that many well-established clinical characteristics that are reliable predictors in prognosis and survival (i.e. tumor stage, size, patient & age) end up being unreliable in a response prediction setting. Thus, the propose is to use a multitude of genomic factors to tackle this problem in hope that this alternative source of information can lead to better prediction results.

Cristian Tomasetti (mentor Professor Giovanni Parmigiani, HSPH/DFCI) is a Postdoctoral Fellow in the Department of Biostatistics at the Harvard School of Public Health and the Dana-Farber Cancer Institute. Dr. Tomasetti received his PhD in Applied Mathematics and a MA in Mathematics, from the University of Maryland, College Park. He will be moving to a faculty position in the department of Oncology and Biostatistics at Johns Hopkins University this fall. The focus of Dr. Tomasetti's work is on the development of mathematical models of cancer evolution, development of drug resistance and cell stem dynamics. During the past two years he has been provided with the opportunity to work on cancer genetics, a fascinating new direction, and to statistically analyze various sequencing datasets.

His main research accomplishments during the time funded by the cancer training grant T32 are summarized below.

Somatic mutations and cancer genome sequencing studies

Today's fast developments in sequencing technologies are creating an ever-increasing amount of data freely available for analysis (e.g. the Cancer Genome Atlas - TCGA). The combination of mathematical modeling and statistical analysis of sequencing data is critical for increasing our understanding of cancer's evolutionary process, enabling us to individuate the key players of this disease. Dr. Tomasetti has developed a stochastic model that provides a general framework for understanding the process of accumulation of somatic mutations in cancer. The model provides a new way to estimate in-vivo tissue-specific somatic mutation rates.

Drug resistance in cancer

By using a mixed-model approach where differential equations were used in combination with branching processes, Dr. Tomasetti has been able to include the complex dynamics of cancer stem cells into a model of drug resistance development. With this result we have highlighted the role that the stem cells' mode of division plays in the dynamics of cancer drug resistance. He has also extended these analytical results, via both a semi-deterministic approach as well as a fully stochastic one, by replacing the standard exponential growth with a more realistic class of tumor growth curves.

With various collaborators at Dana-Farber Cancer Institute and University of Texas MD Anderson Cancer Center, Dr. Tomasetti's theoretical work has been successfully applied to predict drug resistance in gastrointestinal stromal tumors and CML. We are able, for the first time, to provide evidence supporting the idea that mutations responsible for resistance in CML and GIST likely pre-date the start of the treatment. Finally, Dr. Tomasetti has formulated a new hypothesis on chronic myeloid leukemia stem cells and their resistance to the treatment.