Syllabus - 02701 - Current Topics in Computational Biology
                                        Spring 2017
                                      Carl Kingsford

GOALS

    * Read a broad spectrum of papers to give everyone some familiarity with
      many areas of computational biology.

    * Practice and learn to make high quality scientific presentations.

    * Learn how to critically read scientific papers.

    * Learn how to give and take constructive feedback to peers.

SCHEDULE

Wednesdays, 3:30-4:50 in Wean 8427

1   Jan 18  Introduction, goals of the course
2   Jan 25  
3   Feb 1   -- NO CLASS
4   Feb 8
5   Feb 15
6   Feb 22
7   Mar 1
8   Mar 8
9   Mar 22
10  Mar 29
11  Apr 5
12  Apr 12
13  Apr 19
14  Apr 26
15  May 3

COURSEWORK AND GRADING

The main work of this course is to read the assigned papers, answer some
questions about the papers, and then to participate in a discussion of the
papers during class. In addition, 2 (sometimes 1) people will create a
presentation on the paper each week. Practicing how to effectively present
research is another significant part of the course.

The presentations should be 25 minutes long. We will strictly enforce that
limit.

Each class, following the presentation, we will have a discussion of the paper
followed by a discussion of the presentation. You are expected to contribute to
both.

I will provide a written critique of each presentation (based on a form I will
provide).

In addition, each week, we will distribute a short questionnaire about the
paper. These questions will help to guide the discussion following the
presentation. The questions will likely change from week to week.

Your grade will be in equal parts based on:

    * Participation (attendance + discussion): 33%
    * Presentation quality: 33%
    * Questionnaire quality: 33% 

In addition, failing any category will lead to a non-passing grade.

We will likely use the extra 2 weeks of class time for other discussions
related to paper reading / writing / presenting.

EXAMPLE QUESTIONNAIRE QUESTIONS

Each week, we'll email you the questions that you are to answer that week.
These questions will be used to start the discussion.  Below are a few example
questions just to give you a sense of the type of questions that might appear.


    1. What are the biggest strengths of this paper?

    2. What are the biggest weaknesses?

    3. Give a follow up experiment or analysis that could be the basis of a
    future paper:

    4. Do you think this paper is in the right venue? (too good? too bad? wrong
    audience?) Why or why not?

    5. Write a better title for the paper.

    6. What is the worst written paragraph in the paper and why?

GUIDELINES FOR PRESENTATIONS

The goal of these presentations is to save your audience time.  If you take 30
minutes to present a paper that they could have themselves read in 30 minutes
you have not succeeded (and because of the imperfect of fidelity of
communication, you have likely failed). This is the most important rule of
presentation.

Strive to make your presentation efficient by addressing the paper's main
points, main ideas, and main weaknesses.  

Do not present the paper in the order in which it is written. Instead, present
in layers: Present the main point first: why did the authors conduct this
study? What have they learned? If the audience were to remember one thing from
the paper, what would it be? Look in the abstract, end of the introduction, and
the conclusion for hints about what the authors think this is. (1--5 minutes)
This part should be crystal clear: if everyone left the talk after this point,
they should still be able to fake their way through a 1 minute conversation
about the paper.

Early on in the presentation, you should describe the broad outline of how the
experiments were conducted or the basic idea of the algorithm or the main point
of the proof.  Be brief, assume your audience knows the basic techniques.
Examples: "They used a Bayesian network, trained on these examples, to create a
classifier for graphs of size < 10"; "They used linear programming find the
parameters that caused the model to grow networks that best matched the
training set of graphs"; "They combined experiments in directed evolution with
NMR". This gives your audience a mental roadmap and some context.

Next, present some details: the depth of detail should depend on (1) the
importance of the topic to the paper and (2) whether it would be better for
people to read a part on their own. Not everything is appropriate for a
presentation.  This could be the bulk of your presentation, time-wise, though
it is likely not the most important part in terms of creating a good
presentation.

Finally, wrap up by repeating *very briefly* the main points of the paper  and
then end with some questions, suggestions for follow ups, room for
improvements, crazy ideas the paper inspired, etc. (~ 2 minutes)