 |
|
Posters
|
|
Abstract |
|
|
|
|
|
|
|
|
|
Performance Factors in Grid Processing of Phototaxis Data
|
|
|
|
|
|
|
|
Claire Rinehart1, Daniel Fitzpatrick1
|
|
|
|
|
|
|
|
A phototaxis instrument has been set up, with 60 sample channels, to monitor the
movement of Chlamydomonas towards a light source. The instrument is controlled by
LabView and the number of illumination periods as well as their frequency can be
entered at the beginning of each run. Typically, between 200 and 1200 illumination
periods are used. During each illumination period 16 density measurements are recorded
creating a data matrix that is up to 60 x 1200 x 16. For each sample channel, rhythmic
curves are fit to each of the 16 samples taken over all of the illumination periods
to give 60 X 16 fit curves. From each curve estimates of period and phase are calculated.
These calculations have been set up using program Mathematica. Processing all of
these data can take over 39 hours on a single processor. We have used Mathematica’s
Parallel Computing Toolkit to take advantage of six-node dual-processor bioinformatics
cluster. This package uses our 12 cluster processors and was able to reduce processing
time to a little over 4 hours. In this study we report the factors affecting the
performance of the Grid based system.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1Department of Biology, Western Kentucky University, Bowling Green, KY,
USA |
|
|
|
|
|
|
|
|
