|Department of Biostatistics
Ryals Public Health Bldg, 317
University of Alabama at Birmingham
Birmingham, AL 35294
Phone: (205) 996-4154
Fax: (205) 975-2540
Dr. Cui received her Ph.D in Genetics at Iowa State University in 2001. She then took her 3-year postdoctoral training in statistical genetics at the Jackson Laboratory. Dr. Cui joined SSG in Aug 2004 as an Assistant Professor with a joint appointment in the Biostatistics Department and the Department of Medicine (Division of Genetic and Translational Medicine). Her research interest includes statistical genetics/genomics, epigenomics, and high throughput experimental design and data analyses.
The conventional supervised analysis of high dimensional expression data (microarras, proteomics, RNA-seq etc) often identifies differentially expressed genes by rejecting the null hypothesis of equality at each gene. The genes that are not significantly different are often discarded as uninteresting. However, when the goal of the experiment is to discover genes, pathways, or networks that are consistent (not changing), such as evaluating model systems and seeking common mechanisms between related diseases, the conventional "testing for difference" framework does not provide a valid test directly answering these questions. Many people have the misconception that genes failing to be categorized as differentially expressed according to the statistical tests for differential expression must be equivalently expressed. This misconception fails to recognize that the reason that a gene is not categorized as a differentially expressed gene is most likely due to the lack of power caused by high noise level or lack of replication in the experimental design, or both. The lack of statistical evidence of differential expression does not provide evidence of equivalence in gene expression. In this project, we are trying to develop, evaluate, and apply sound equivalence testing methods for identifying equivalently expressed and equivalently changed genes, pathways, and networks. Our current application of this method is in evaluating the consistency of gene expression change between primary human brain tumor (GBM) and their xenografts in mice, which is critical for the successful transfer of drugs tested on xenografts to human patients.
Next Generation Sequencing is a booming technology that has many applications in Biological Sciences, such as RNA-seq for quantifying transcript expression, ChIP-seq for interrogating protein binding sites in the genome, exome-seq for sequencing all the exons in the genome, as well as whole genome sequencing and whole genome DNA-methylation sequencing etc. For this technology, we are mainly working on the experimental design and data analysis of RNA-seq data and ChIP-seq data.
We were funded for the organization of a conference focusing on the Statistical Analysis of NGS data.
Statistical Analyses for Next Generation Sequencing
Sept 26-27, 2011
Modifications to DNA or chromatin that affects the gene expression but without altering the DNA sequence are often referred to epigenetics. The whole collection of this type of modifications is referred as epigenomics. Since epigenetic modification can be results of environment factors, it is often considered to be the bridge between genetics and environment in human disease. We are interested in the data analysis of epigenomics data as well as study the role of epigenetic alterations in human cardiovascular and neural disease. The technologies we work with are ChIP-chip, ChIP-seq, and DNA methylation microarrays.
BST 675 - Introduction to Statistical Genetics (offered in the spring of odd years)
This is an elective course in the Biostatistics Department. It is to introduce students to basic concepts in population genetics, genetic epidemiology analysis, Mendelian laws of inheritance, heritability, test cross, linkage analysis, QTL analysis, human linkage and human association methods for discrete and quantitative traits. Students are expected to critically read and present current statistical research papers and to write a mini-review summarizing an area of statistical genetics research that is of interest to the student.
BST 676 – Genomics Data Analysis (offered in the spring of even years)
This is an elective course in the Biostatistics Department. The purpose of this course is to teach graduate students statistical methods that underlie the analysis of data generated by high throughput genomic technologies, as well as issues in the experimental design and implementation of these technologies. High throughput technologies that are covered include: microarrays, proteomics, and second generation sequencing. Many of real data analysis projects will be provided as the class projects. After the students go through each step of the analysis and finish the project, they are expected to take on new but similar data analysis projects.