Math/Stat 750
Introduction to Interdisciplinary Consulting
Fall 2003 - Matthias K. Gobbert and Nagaraj K. Neerchal
Class Presentations of the Consulting Projects
This page can be reached via the homepage
of the Scientific Computing and Statistical Data Analysis Lab at
http://www.math.umbc.edu/~gobbert/scsdal.
Program for Day 1
Tuesday, December 02, 2003, 10:00 a.m., MP 401
Session chair: Matthias K. Gobbert
-
10:05-10:15
An Algorithm for Locating Amino Acid Residues in Proteins
Samuel G. Webster
Client: Dr. Mauricio M. Bustos,
Department of Biological Sciences,
UMBC
Proteins are molecules composed of linear sequences of amino acids.
Each protein possesses its own unique, intricate, three-dimensional structure
that determines its functionality. Very often, amino acids that
are far apart in the linear sequence are found next
to each other in the protein. The goal of molecular
biologists is to correctly identify these amino acids and then
alter the structure of the protein. I have designed an algorithm that
extracts a 3-D volume element from a protein and returns all amino acids
that lie within the element. Additionally, the linear sequence with the
highlighted amino acids is returned.
Project mentor: Matthias K. Gobbert
-
10:20-10:30
Analysis of Nursing Behavior in Mother/Calf Dolphin Pairs
Karen L. Osborne
Client: T. David Schofield, Manager,
Ocean Health Programs/MARP,
National Aquarium in Baltimore
It is suspected that there is a difference in nursing and other
care-giving behaviors between experienced dolphin mothers and
inexperienced, e.g., first-time dolphin mothers. Quantifiable differences
in the occurrence of certain behaviors can be identified as playing an
important role in the survivability of the calves. Behavioral data was
provided on three dolphins and their calves, for a 10-week time period.
The assumption is that the frequency data provided follow a Poisson
distribution, and that these counts are potentially influenced by dolphin
and time. A Poisson regression model was fit to the count data using these
variables, and appropriate tests were developed to determine if the
differences were significant.
Project mentor: Nagaraj K. Neerchal
-
10:35-10:55
Statistical Analysis of Proteomics Data
Ravi Siddani and Alex Sverdlov
Client: Dr. Brian P. Bradley,
Department of Biological Sciences,
UMBC
An important application of proteomics is comparison of 2D gel images of
protein mixtures taken from several treatment groups. These images can be
coded as binary 30X30 matrices, with 1 indicating the presence, and 0 the
absence, respectively, of a protein in a gel. Two statistical methods, a
permutation test, and a cluster analysis have been implemented to see
whether gels taken from 6 treatment groups have different structures.
Distributions of permutation test statistics and the corresponding
p-values were obtained for each pair of treatments under consideration.
Clusters of gels showing the differences between treatment groups were
obtained. The results are consistent for the two considered methods.
Project mentor: Nagaraj K. Neerchal
Program for Day 2
Thursday, December 04, 2003, 10:00 a.m., MP 401
Session chair: Nagaraj K. Neerchal
-
10:05-10:25
Lameness Index in Dairy Cattle
Minglei Liu and Yanping Wu
Client: Dr. Uri Tasch,
Department of Mechanical Engineering,
UMBC
A team led by Dr. Tasch developed a Reaction Force Detection
(RFD) system to predict the lameness of cows, which is a very important
issue for the dairy industry. The objective of this project is to
evaluate the effectiveness of the RFD. In the project, a cutoff point is
chosen for the predicted lameness index which minimizes
misclassifications, and the various associations among the lameness and
ralted variables are demonstrated by variety of plots and tables.
Project mentor: Nagaraj K. Neerchal
-
10:30-10:40
Using a Fourier Method to Solve a Convection-Diffusion Equation
Zhibin Sun
Client: Dr. Andrew Tangborn,
Global Modeling and Assimilation Office,
NASA Goddard Space Flight Center
For a particular kind of partial differential equation, the
convection-diffusion equation with periodic conditions, we can use a
Fourier method, which is implemented by FFT, to transform it into a system
of ordinary differential equations. These are solved using the
Crank-Nicolson scheme for time-stepping. We will see that this Fourier
method works well for certain problems, whose frequency is in the range of
the Fourier expansion. The results of experiments show that the method is
effective.
Project mentor: Matthias K. Gobbert
-
10:45-10:55
A Finite Difference Solution of a One-Dimensional Non-Linear
Reaction-Diffusion System with a Fast Reaction
Ana Maria Soane
Client: Dr. Thomas I. Seidman,
Department of Mathematics and Statistics,
UMBC
A chemical process involving three reactive species is modeled by a
non-linear system of three reaction-diffusion equations in one spatial
dimension. The problem is challenging numerically, because one
reaction is much faster than the other one.
Mathematical analysis and numerical results exist for the steady-state system.
The goal of this project is to solve the
time-dependent system numerically, to confirm the intuition on the
system's behavior. After discretizing in space
using finite differences, implicit time stepping is used. Results both
for the steady-state problem and for the transient problem will be shown.
Project mentor: Matthias K. Gobbert
Copyright © 2003 by Matthias K. Gobbert and Nagaraj K. Neerchal.
All Rights Reserved.
This page version 1.2, November 2003.