Accession Number : ADA523910


Title :   Parallel Implementation of a Bioinformatics Pipeline for the Design of Pathogen Diagnostic Assays


Descriptive Note : Conference paper


Corporate Author : ARMY MEDICAL RESEARCH AND MATERIEL COMMAND FORT DETRICK MD TELEMEDICINE AND ADVANCED TECH RESEARCH CENTER


Personal Author(s) : Satye, Ravi V ; Kumar, Kamal ; Zavaljevski, Nela ; Reifman, Jaques


Full Text : http://www.dtic.mil/dtic/tr/fulltext/u2/a523910.pdf


Report Date : 19 Jun 2009


Pagination or Media Count : 8


Abstract : The genomes of hundreds of pathogens and their near neighbors are now available and many more are being sequenced. With the availability of this genome information, sequence-based pathogen identification has become an increasingly important tool for clinical diagnostics and environmental monitoring of biological threat agents. Chief among sequence-based identification tools are DNA microarrays, which have the ability to test for thousands of pathogens in a single diagnostic test. The design of microarray diagnostic assays involves the identification of short DNA sequences unique to a pathogen or groups of pathogens where these unique sequences, or ?fingerprints? (also referred to as probes) are used to identify the pathogens. To design pathogen fingerprints, we developed TOFI (Tool for Oligonucleotide Fingerprint Identification), a high performance computing software pipeline that designs microarray probes for multiple related pathogens in a single run. The TOFI pipeline is extremely efficient in designing microarray fingerprints for multiple pathogens. Parallel implementation of computationally expensive specificity analysis of the designed fingerprints drastically reduces the overall execution time of the software. Comprehensive performance analysis shows that TOFI achieves super-linear speedup for up to 74 processors. A Web-based user interface developed using the User Interface Toolkit, provides easy access to the pipeline. Using 74 processors, TOFI took approximately nine hours to design 5,015 in-silico probes for eight Burkholderia genomes with a combined size of more than 50 million base pairs. Experimental validation of these probes with various Burkholderia genomes showed that nearly 80% of the designed fingerprints identify the intended targets.


Descriptors :   *BIOASSAY , *BIOTECHNOLOGY , *BIOLOGICAL AGENTS , GENOME , MONITORING , SYMPOSIA , DIAGNOSIS(GENERAL)


Subject Categories : Biochemistry
      Biomedical Instrumentation and Bioengineering


Distribution Statement : APPROVED FOR PUBLIC RELEASE