Discovering Significant Pathways of Gene Regulation

Researcher: Laura Poulton Mentor: Serdar Bozdag

Summary:

Recent advancements in biotechnology have made it possible to generate vast amounts of gene expression data for thousands of organisms. The collection of high-throughput gene expression data allows computational biologists to develop algorithms to reverse engineer the underlying gene regulatory network (GRN) of cells from their gene expression data. Among several software tools to reverse engineer GRNs, FastMEDUSA is a powerful tool. FastMEDUSA builds a model represented by an alternating decision tree (ADT) that predicts the potential regulators of genes.

We hypothesize that if there are significantly overrepresented branches in the ADT of FastMEDUSA, they could be biologically important pathways for gene regulation. In this project, we analyze the ADT built by FastMEDUSA to compute significantly overrepresented branches. We compute p-value for each branch based on results on randomly generated ADTs. For validation, we check public databases and literature to verify if genes in these branches have been reported as pathways of gene regulation.

Students will have the opportunity to work on graph theory, statistics, and biological databases to answer some high-impact biological questions.

Goals:

• Using an already existing computational model for predicting gene expression based on the presence of microRNA and other transcription factors, contribute to a list of known gene regulators and co-regulators.
• Create an R package from the current gene expression prediction model that can be downloaded as a library.

Weekly Milestones:

Week	Description	Status
Week 1	Become familiar with the R Programming language	Complete
Weeks 2 and 3	Become familiar with general molecular biology concepts Conduct a literature search, reading through research papers on previous work in this area Become familiar with the specific R libraries used most often in bioinformatics research	Complete Complete Complete
Week 4	Download mRNA expression, miRNA expression, copy number alteration, and DNA methylation data from TCGA Reformat data sets so that they are easily readable by the applicable R functions and packages Present findings from literature search	Complete Complete Complete
Week 5	Present on REU work completed so far Process TCGA data for differential expression analysis Download and preprocess putative miRNA - mRNA interaction data from miRTarBase and TargetScan databases	Complete Complete Complete
Week 6	Run differential expression analysis Download and preprocess putative TF - mRNA interaction data several databases Filter putative interaction data using differential expression analysis results Create correlation matrices for mRNA and miRNA	Complete Complete Complete Complete
Week 7	Filter regulator - target gene interactions by pvalue and correlation Identify hub genes in Prepare first draft of final research talk	Complete Complete Complete
Week 8	Run clustering algorithm Create data structure to store information on target genes Write first draft of final paper	Complete Complete Complete
Week 9	Perform enrichment analysis on clusters Analyze extracted miRNA - TF - target gene modules Make research poster	Complete In progress Complete
Week 10	Perform survival analysis (if time allows) Present at poster session Prepare and present final research talk Finish final research paper	In progress Complete Complete Complete