|
March-April 2004
Home Page
E-Notes Home Page |
 |
|
Playing in the Sandbox
Research Computing with MATLAB
"It was the best decision I ever made," said Dr. Geoffrey Goodhill regarding his adoption of the research computing and application development software MATLAB.
Dr. Goodhill, an associate professor and researcher with Georgetown University Medical Center's Department of Neuroscience, was a featured speaker at the February 12 seminar "Using MATLAB in the Life Sciences". Sponsored by Advanced Research Computing, the in-depth seminar brought MATLAB Educational Sales Representative Angelo D'Amato, Application Engineer Rob Bemis, and senior Biotech Development Manager Rob Henson to Georgetown University.
Dr. Goodhill and his colleagues in the Computational Neuroscience Laboratory study the development of neurons in the primary visual cortex, aiming to uncover the patterns of neural formation by studying the visual cortex at various stages of development. To extrapolate patterns from large volumes of data, Dr. Goodhill uses MATLAB to convert numbers to graphical models that show at a glance the differences in visual cortex structure at various stages of development. By using MATLAB, Dr. Goodhill can easily visualize and simulate a huge variety of complex scenarios...and most importantly, do it quickly.
MATLAB combines data analysis, graphing, and algorithm and application development into one environment. The main interface incorporates a command-line window for programming in a variety of languages such as C, C++, Java, and Perl. "Toolboxes", discipline-specific add-ons, can be plugged into MATLAB to add functionality. Bioinformatics, Neural Networks, Image Processing, Signal Processing, Statistics, and Genetic Algorithms are a few examples of MATLAB's more than 40 available toolboxes (not counting the many toolboxes made by third-party vendors). For further customization, MATLAB is built on open-source architecture so that researchers can modify any aspect of the program to fit their needs. "The whole idea is that you don't have to break stride," explained Bemis.
Bemis took the audience on "a MATLAB test drive", demonstrating its Data Import Wizard, Plotting and Statistics tools, Image Processing toolbox, Genetic Algorithms toolbox, and several other functions. Seminar attendees learned that MATLAB is capable of importing almost any form of data, (including tab- or comma-delimited text files, binary formats, spreadsheets, databases, and even image, sound, and video), plotting data visually, and applying saved algorithms for batch processing. Also demonstrated were graphing features that allow users to instantly change the level of detail displayed in a visual map, its color scheme, or even rotate it on a 3-D axis. Bemis called MATLAB an "interpretive environment" because it allows you to "play" with data, seeing the results of complex calculations almost instantly. "This is one reason why it shows up in so many different departments on campuses", Bemis said, "A lot of people use MATLAB as a sandbox, to play around with their data, get different ideas."
This point was reaffirmed later by Henson, who said, "MATLAB is great for algorithm development. Plus you get the instant gratification of the graphics." Henson, MATLAB'S lead biotechnology developer, then introduced his "baby", the new Bioinformatics toolbox. Engineered in conjunction with bioinformatics researchers, the toolbox features algorithms specialized to bioinformatics, gene expression analysis tools, microarray normalization and visualization functions, and other capabilities. To demonstrate, Henson downloaded an entry from Georgetown's Protein Information Resource online database directly into MATLAB and with one command plotted the protein's amino acid composition on a graph.
ARC Director Steve Moore believes that adopting MATLAB widely at Georgetown will increase researchers' productivity, enhance their competitiveness for grants, and augment the University's reputation as a research institution. Henson and other MATLAB software engineers are working with Georgetown researchers and ARC to develop a parallel processing toolbox; the seminar helped them begin a similarly close relationship with the Medical Center's life sciences researchers. With positive reactions from attendees, Moore thinks the presentation successfully furthered one of ARC's goals--supporting scientists through informative seminars in the research computing space--and, as Moore said "giving a fair hearing to really cool tools." |
|
