PROJECTS GUIDE FALL 2007


DELIVERABLES

You will need to deliver the following:
  • Monday October 29: a project plan.
  • December 3-4: a 15 minutes project presentation and slides.
  • Monday December 17: A project report.
    All documentations should be submitted electronically (.pdf, .doc, .ppt).

    These will be further explained below.

    HOW TO PURSUE A RESEARCH PROJECT

    The course project is an opportunity to pursue a mini research project and get your feet wet. Here are some useful guidelines how to proceed.

    Research starts with a question and ends with a result. The first step in pursuing research is to formulate both, the question and the expected results in as precise terms as you can.

    Consider for example a project intended to compare certain multiple sequences alignment (MSA) algorithms, say A,B,C. The questions underlying the project may be loosely stated as "how well do the algorithms A,B,C perform the MSA task?"; this question can lead to more precise questions such as "How can the quality of an MSA be measured?", or, "what is the time and space complexity of algorithms A,B,C?". Each of these questions already defines the nature of the potential results. The first question requires a mechanism to best measure the "quality of an MSA" -- e.g., in terms of its biological meaningfulness; such mechanism and the analysis to establish its validity (e.g., applying it to biologically known alignments) constitute potential research results. Similarly, the second question can result in complexity analysis of the respective algorithms. Furthermore, it is possible to establish the novelty and potential significance of such results relative to previous knowledge.

    Therefore, it is beast to start your project by (a) formulating the questions you will investigate and the results you expect in precise terms; and (b) establishing the degree to which such results would be novel and significant.

    Research Results Generally Fall Into Four Categories:

    Surprisingly as it may sound, it is often useful to define first the target results. Edison is said to have started his research projects by first writing patent claims describing the novel elements of his invention, then planning and experimenting with alternative techniques to accomplish the result. Starting with the results simplifies the task because results fall generally into four categories:

  • An Algorithm/Mechanism: e.g., a new algorithm for multiple sequence alignment; a Hidden Markov Model (HMM) to compute active protein sites from sequence data; a mechanism to evaluate the biological meaningfulness of an MSA; a support vector machine (SVM) to learn and classify sequence motifs.
  • Analysis Of An Algorithm/Mechanism:e.g., complexity analysis of an MSA algorithm; analysis of microarray or flow cytometry data to extract regulatory relationships among genes; analysis of dynamic behavior of a regulatory mechanism.
  • A System Design: e.g., an indexing system to accelerate sequence database searches; a language to construct and analyze biological network models; hardware to accelerate protein conformation computations.
  • A Theory:e.g., a mathematical theory to compute Boolean regulatory network models from temporal expression data; a mathematical theory to compute pathways in metabolic networks.

    What Makes A Result Valuable?

    A result is valuable to the extent that (a) it is novel and not obvious; and (b) it is of broad applicability to improve the state of the art. These guidelines can help you assess the value of research results in publications that you read, as well as the potential value of the results targeted by your project (needless to say that the value of the results your project accomplishes are constrained by the time and resources available).

    THE PROJECT PLAN

    The Project Plan shall include the following sections:
  • Specification of the question you plan to attack and the results you expect to accomplish.
  • Background and prior work: summarize briefly previous research and identify the novelty and potential value of the results you plan to accomplish
  • Work plan Schedule: describe your work plan in terms of schedule of milestones and activities.

    The project plan is an essential tool in pursuing a research project. The first section aims to help you focus on specific, precisely defined, results. The second section aims to help you understand the body of existing knowledge and key relevant results in the project area. To prepare this second section you can use the class material as well as key publications in the project area. The third section aims to help you set concrete schedule of work and intermediate milestones to best ascertain your success. The project plan document will also serve as the basis from which you derive the project report.

    There are several types of projects you can pursue.

    A Comparative Survey

    A comparative survey may be a useful beginning of a research and can accomplish very good course project results. A comparative survey of an area can (a) identify the overall challenges and questions that the area is trying to answer; (b) compare the different results of the area; and (c) identify open questions not yet resolved by these results.

    Your own primary results could include comparative performance evaluation of key techniques in the area, or analyses of intrinsic bounds on these techniques. For example, a survey of techniques to extract regulatory network structure from microarray measurements can establish metrics to evaluate the resulting network models, compare the performance of several such techniques relative to a particular regulatory network (e.g., the Yeast's regulation of mating, metabolism, or response to heat shock).

    A New Algorithm/Mechanism

    You can investigate new algorithms or mechanisms to resolve biocomputing questions. For example, you can investigate indexing and algorithms to accelerate searches of sequence or microarrays databases; SVM architectures to detect various components of a protein (e.g., active parts) from its sequence structure; HMM architectures to determine protein folds; or a Bayesian Network to analyze signaling cascade structures from flow cytometry methods.

    Budget Your Time & Schedule Your Progress

    A course project needs to be completed within a tightly constrained time and resources. This means that you must: (a) carefully select the target results to be ones that you can accomplish; (b) prepare a careful project plan of tasks, milestones and deliverable sub-results that will accomplish your project goal.

    THE PROJECT REPORT

    The Project Plan shall include the following sections:
  • Abstract: briefly summarizing the key results of your project and their novelty relative to prior work
  • Introduction: describe the questions address by your research and their relationships to previous work.
  • Results: describe the key results of your project and establish their validity.
  • Prior Work: summarize briefly previous research and identify the novelty and potential value of your results.
  • Conclusions: briefly evaluate future research that may be suggested by your project

    If you pursue a survey, your report should expand the Prior Work section and relocate it before the Results section. The Prior Work section should provide more extensive description of the different approaches and key results of the survey area. The Results section should be replaced by a Comparative Evaluation section where you establish quantified comparison metrics and provide comparative performance analysis/measurements of these techniques applied to selected scenarios; this section should explain the key findings of these comparisons. The Conclusions section should focus on identifying key open questions for future research of the area.

    THE PROJECT PRESENTATION

    I will conduct individual project presentation sessions to help maximize the project success. The Project Presentation shall provide a 15 minutes overview of the project results. It should describe the questions explored by the project, establish the key results obtained by the project and evaluate the novelty and significance of these results. You can use the feedback from the presentation session to improve your final project report. I will provide a .ppt template for the project presentation slides.


    Project References & Resources


    Please consult the Older Project Resources Guide as well as the newer Project Resources Guide which is under construction.

    Project Presentation Template


    TBD