|
CS Home
Dept Info/Contacts
People
Research
Events
Courses
Undergrad Programs:
Computer Science
Informatics
Graduate Program
Prospective Students
Faculty Hiring
Employment
Resources
Help Lab Hours
Student Groups
Support the Department: Weeg Professorship
|
|
Profile-based alignment algorithms for protein three-dimensional structure
prediction
Friday, April 10, 2009
4:00-5:00pm, 140 SH
Abstract
The three-dimensional structure of a protein is key to understanding its
function. However, the protein structure determination by X-ray
crystallography or NMR spectroscopy is time- and resource-consuming
process. Various computational methods have been proposed to address the
drawbacks of the experimental techniques. These methods predict (rather
than accurately determine) the structure of a protein from its primary
sequence. For example, the structure of a protein can be approximated
based upon the relationship (alignment) between the protein's primary
sequence and the amino acid sequence of another protein of known structure
(template). This approach, known as comparative modeling (or homology
modeling), is currently the most accurate protein structure prediction
method, but its usefulness is limited by the quality of the alignment
between the query and the template protein. A number of computational
techniques have been developed to overcome this limitation, including
profile-based algorithms for sequence alignment.
We study the relationship between the accuracy of a profile-profile
alignment algorithm and the size of the sequence profiles, where the
profile size is defined as the average number of different residue types
observed at profile's positions. Using an in-house alignment method
UNI-FOLD, we demonstrate that optimizing the size of sequence profiles can
dramatically improve the inference of weak relationships between protein
sequences. Our method performs particularly well at the SCOP superfamily
level, recognizing over 69% of superfamily relationships in the Lindahl
benchmark for fold recognition.
Short Biography:
Aleksandar Poleksic, Ph.D., is an Assistant Professor in the Department of
Computer Science at the University of Northern Iowa. His training and
interests span the fields of computational biology and bioinformatics,
with a strong emphasis on novel algorithm and application development.
Prior to joining UNI, Dr. Poleksic was a Senior Scientist at
Eidogen-Sertanty, Inc., where he helped develop Eidogen's computational
drug discovery platform that integrates numerous algorithms in the area of
protein structure determination and analysis. Dr. Poleksic was recently
granted a United States patent for alignment algorithms and methodologies
for rapid protein homology detection.
|
