|
Syllabus
Assignments
Lab
Worksheets
548 Resources
Web
Resources
Term
Project
Speaker
Schedule
Outside
Seminars
Contact
Instructor |
Instructor: Dan Burns
Office: 5834 East Hall
Phone: 763-0152
E-mail: dburns@umich.edu
Under Construction!
This page will provide a convenient
access to the class events which are being scheduled, the assignments
and group rosters, and a convenient set of links to (online) papers and
web sites which will be of use in the course. For your convenience, here
is the first
day handout summarizing the course, as well as the
course announcement which gives a bit more syllabus detail. Note,
however, that I am thinking of modifying a few things over the course
of the term due to recent developments.
Problem set assignments will be
available below (follow link in the left side bar). First problems sets will be due Wednesday, February 4.
There will be no examinations
in the class. There will be a final project which will consist of your
studying on a particular subject and making a presentation to the class.
This will be a twenty to twenty-five minute Power Point presentation,
and will be done in teams of two. There will be a page of suggested topics,
though you will be free to choose a topic of your own (it must be approved,
however).
Link here
to the page for the final project, including suggested topics,
etc.
Schedule of Readings and
Detailed Syllabus:
Click on a section subject for
a relevant link, if available. DE means Durbin, Eddy, et al., ``Biological
Sequence Analysis"
| |
|
|
|
March 3-7,
2003
|
Read: Durbin and Eddy,
Chapters 7 & 8
(as much as possible) |
Phylogeny, especially
for protein families. |
Many approaches to phylogeny,
which is, again, a
computationally hard problem. |
March 7-10,
2003 |
Kahn, Qian & Goldstein 2000,
Qian, Goldstein 2002. |
Phylogeny, especially
for protein families, especially one use for making multiple sequence
alignments more accurate. (The preprint links -- right -- are more
directly relevant than the reprint links -- left.) |
Tree
based HMM's for m.s.a.
and classification
of GPCR's
(= G-protein coupled receptors). |
March
12-19
(March 14 canceled),
2003
|
Read: DE, Chap. 7 (parsimony);
Chap. 8 (ML);
HMM: Felsenstein-Churchill 1996,
Mitchison-Durbin, 1995 (not linkable; will be distributed in class). |
Phylogeny: ML; Parsimony;
HMM's in phylogeny.
|
Parsimony most used method for tree estimation; HMM's vary substitution rates across sequence positions.
|
March
21-24,
2003
|
Reference: Brian Ripley, ``Pattern Recognition and Neural Networks", Camb.UP (1995),
Ch. 5: Feedforward Neural Nets.
|
Basics of NN's:
Feedforward nets,
supervised training, backpropagation algorithm;
gradient descent minimization.
|
The ``vanilla" settings for NN's; the complete literature is vast;
few rigorous arguments, very heuristic field.
|
March
26-28,
2003
|
Neural nets in promoter recognition: NNPP;
M. Reese, Comps. & Chem., 26 (1998) 51-56.
|
Time delay NN's;
application to eukaryotic promoter site recognition.
|
Typical use of
NN for pattern recognition, with modification to allow for
flexible location for recognition of the ``same" signal.
|
March
31, 2003
|
Probabilistic
version of promoter recognition: McPromoter.
Ohler et al., 1999.
|
Interpolated
Markov chains;
application to eukaryotic promoter site recognition.
|
Use of
higher order Markov chains for pattern recognition, with modification to allow for
flexible use of available data: weighted use of shorter and
longer context sequences, with (non-probabilistically
enforced) weighting of more commonly occuring context sequences.
|
April
2 - 4,
2003
|
Improvements in McPromoter;
Ohler et al., 2001.
|
Incorporating
biophysical properties of sequences.
|
Ohler's
extension of McPromoter to include DNA physics;
intro to duplex stress and gene promoters (after Benham et al.).
|
April
4 - 14,
2003
|
Term Project
Presentations:
Good luck!
|
Great
variety of topics.
|
Visitors
welcome:
schedule of speakers.
|
April
14 - 17,
2003
|
End of Term:
Sonnhammer
et al., 1998.
Martelli et al. (2002).
|
Trans-membrane
proteins: recognizing helices and beta barrels.
|
Using HMM's
for structural feature recognition.
(Papers taken from the suggested project topics.)
|
|
