Prof. Ron Shamir
School of Computer Science
Tel Aviv University
March 2006
Seminar on Algorithmic Problems in Computational
Biology
Guidelines
Your Talk: What, How and Why:
Give clear definitions of the terms you are using,
and of the problems you will study. Include examples to add intuition. You
may go over these fast if the background and terms were defined and used
in previous lectures, but do not skip them. You should not "lose" anyone
in the audience at this stage!
Provide motivation for the problem: why is it interesting?
relevant?
If you are assigned a paper with more material than
you can include in your talk (this is almost always the case), you should
make a judicious choice what to include and what to leave out. This choice
shows how well you understand the paper and the context. Obviously all key
results should be stated, but only some of them should be proved. It is of
the utmost importance to include the essence of what is most original and
important in this paper. (This is not always the technically hardest result).
Seminar proof style is very different from that
of first year lecture proofs and from the style of proofs written in a paper.
You should make an effort to convey the main ideas and leave out standard
details. Some level of technicality is essential, but keep it to the most
important and most novel aspects of the paper. Usually proving in detail
1-3 theorems or key lemmas will suffice. For the rest (as well as for the
key theorems that you prove in detail) you should supply intuition on why
the statement is correct or why the algorithm works as stated. This is one
of the hardest challenges, as (alas,) many papers are written in a style
devoid of that intuition.
Plan your time, and make several "dry runs" of your
presentations, timing yourself. Make plans what to skip if you run out of
time so that the talk does not end before you reached the "punch-line". Leave
10-15 minutes to questions and discussion, out of which at least 5 minutes
at the end of the talk.
Try to include something in your presentation that
was not in the original paper: A simpler proof, a stronger result, a variant,
insights. Do not be shy and say explicitly in your presentation that this
is your own contribution!
Many presentations will cover both an algorithmic
part and a more experimental part dealing with real biological data. Keep
a good balance between the two. The main focus should be on the algorithmics.
However, you should fully understand and explain in class the biological
part and the specific choices made in analysis of the real data.
An excellent guide on how to present a theoretical
computer science talk, by Ian Parberry, is available at http://www.eng.unt.edu/~ian/guides/speaker.html.
Note that some presentations in the seminar will have a considerable biology
component, so take these guidelines with a grain of salt.
Consult the lecture notes of my course Algorithms
in Molecular Biology at http://www.cs.tau.ac.il/~rshamir/algmb.html
for the basics of the area, and for additional bio- and bioinfo- literature.
Prepare slides (transparent, for overhead projector, or electronic
for ppt-type presentation). Use the blackboard sparingly to save time and
only for critical points.
If your talk is given two hours, plan for a break in the middle,
and time yourself to end at 12:45 (12:50 including questions) sharp.
There should be a pc and a pc projector in class. If you need
transparencies project for your talk, you should coordinate this at least
a week in advance with person in charge of the building.
Make 15 xerox copies of your slides and bring them to class,
so that everyone can have a copy in front of him/her during the lecture (ask
the workers in the copy room on the first floor of Schreiber Bldg. to do
this with my permission. Larisa can help you if there are obstacles here.)
Do not plan to do the copying on the last hour before the seminar (remember
Murphy's law..).
Active participation in the seminars:
In some seminars, the students are only required to show involvement
in their own talk, and are completely passive in the rest. This makes the
seminar very dull for all involved. This is not what is expected here: You
are expected to read also all the other papers in the seminar and be active
when they are presented: Ask questions, add your insights, etc. The
topic of this seminar is very focused and all papers deal with closely related
problems, which makes it possible to easily understand what other papers
are talking about.
How your grade is determined:
40% understanding of the material
40% presentation of the material
10% good choice of what material to present
10% bonus on originality
-10% if you do not complete on time: Punctuality is crucial!