Plant & Animal Genomes XIII Conference

January 15-19, 2005
Town & Country Convention Center
San Diego, CA

Design And Calibration Of Microarrays As Universal Transcriptomic Environmental Biosensors

Jonas S Almeida¹ , Matthew J Jenny^2,4 , Chen A Yian¹ , Javier Robalino^2,4 , David J Mckillen^2,4 , Paul S Gross^2,4 , Robert W Chapman^3,4

¹  Dept Biostatistics Bioinformatics and Epidemiology, Medical Univ South Carolina, 135 Cannon Street, Charleston, SC 29425, USA

²  Biochemistry & Molecular Biology, Medical Univ South Carolina, 173 Ashley Ave, Charleston SC 29425, USA

³  South Carolina Department of Natural Resources, 331 Ft. Johnson Road, Charleston SC 29412, USA

⁴  Marine Biomedicine & Environmental Sciences Cntr, Hollings Marine Lab, 331 Ft. Johnson Rd., Charleston, SC 29412, USA

Transcriptomic signal of an organism conveys reflects both its genetic makeup, and the environmental factors to which it is responding. However, not only are the intrinsic and external influences convoluted in the transcriptomic response, but it also may vary from one individual to another as a result of genetic variability, circadian rhythms or some pathology such as viral infection. Therefore, using an expression microarray as a biosensor may appear to be an hopeless exercise. On the other hand, the use of multiparametric expression patterns has emerged in the biomedical sciences as one of the most promising routes to early diagnosis and, more recently, prognosis, of systemic diseases such as oncogenesis and autoimmune reaction. At the MarineGenomics.org initiative we are using the same approach to enable the use of sentinel organisms such as shrimp and oyster for environmental quality assessment.
The combination of technologies involved in this initiative include the usual probe selection, microarray design and fabrication but a less usual integration with advanced machine learning algorithms that can infer predictive models directly from a constantly widening and varying training data set. This presentation will focus on the issues of experimental design to produce that data dataset, the computational techniques for calibration and mathematical modeling of the transcriptomic response, and the critical need for a web-based environment integration of data submission, model identification and it usage for predictions based on newly acquired microarray results. A preeminent role in this initiative is played by artificial intelligence techniques.