CSE 527 Computational Biology     [ Home Schedule Handouts Project ]

Project deliverables
  • Project proposal (due 10/17)
  • Midterm report (due 11/17)
  • Final report due (due 12/14)
  • Final poster session (on 12/5)
Evaluation criteria 
  • Presentation
    • How well the final report is written in terms of grammar, organization and clarity of description; and
    • How clearly the student presents the problem, the excellence of his or her analysis and the results in the poster session
  • Results
    • Showing that the proposed approach led to interesting biological findings; or
    • How well the proposed approach works in statistical experiments, compared with baseline methods or state-of-the-art algorithms; or
    • Showing that you made hard efforts to accomplish desired results (e.g., performed many meaningful experiments)
  • Novelty
    • Did the student ask novel biological questions; or
    • Did the student propose novel computational methods
  • A project that satisfies all listed criteria (with 'and' instead of 'or') will get significant bonus points.
Examples of project topics include:
  • Predicting survival time of cancer patients based on their RNA gene expression levels.
  • Predicting sensitivity to chemotherapy drugs based on RNA gene expression level.
    • Goal: Our goal is to develope a prediction system for predicting the drug sensitivity based on the RNA levels of genes.
    • Data: We have two datasets consisting of gene expression data (genes x patients) and drug sensitivity profiles (drugs x patients) from cancer patients. [Gene Expression Data] [Drug Sensitivity Data]
  • Understanding how genes are wired differently in the transcriptional regulatory networks in different subtypes of cancer
    • Goal: Our goal is to understand how differently genes regulate each others’ expression levels in each subtype. One way is to learn the regulatory network (we will cover this in class) in each subtype and interpret how they similar/ different. A different approach is to build a classifier that can predict the subtype of leukemia based on the expression data from a patient, which will enable molecular dianosis of leukemia.
    • Data: We have an expression dataset measuring RNA levels of  20,000 genes from 2096 patients suffering from leukemia. There are 18 sub-types of leukemia and it is important to understand the expression signature that characterizes each subtype of leukemia. [Gene Expression Data] [Classification Labels]
  • Identifying genetic factors for metabolic traits
    • Goal: Our goal is to identify genetic loci that contribute to these important phenotypes and how they interact with gender or smoking status.
    • Data: 
  • Understanding the evolutionary changes in the regulatory network between two yeast species
  • Clustering genes in microarray data
  • Understanding the evolutionary change of transcriptional regulatory networks in yeast.
  • Identifying causal sequence variations for related phenotypes.
  • Detecting genetic interaction from genome-wide association studies data.
  • Detecting epistasis from microarray gene expression data.