The execution of the BIOSAP pipeline is performed through a Java class BioSapThread. This allows multiple executions of the pipeline to occur concurrently on different data. The threads are created within another Java class BioSapper, which runs as a background process checking a specified input directory for new directories containing data. Once a new directory has been located the process launches a new pipeline thread on the data present in that directory. The directories are created and populated through a web interface which gathers the user input necessary to execute a search.

Each execution of the BIOSAP pipeline occurs through the serial execution of the following programs:

1. featurama

   This C++ program searches each bio-sequence in a specified FASTA formatted library specified for valid features according to the algorithm shown in Figure 1. Features that are found to contain duplicates spanning multiple sequences in the library are deleted and the search continues recursively in the sequences in which the duplicate was detected and deleted. Thus, eliminating bias for any bio-sequence and searching each as far as the user set parameters permit. The parameters used for the search are partly based on the parameters used by Primer3 (Rozen & Skaletsy, 1997).

   
   
2. blastall -p blastn

   Running BLAST eliminates all features containing close matches to bio-sequences other than the one for which the feature was chosen.

   
   
3. filterBLAST

   This java application uses a SAX ContentHandler implementation, Jfeaturama, and ISB's own, Robert Hubley's LSAX parser (to be submitted to BioJava), NCBIBlastLexer, to process the raw BLAST output as XML. The BLAST output is filtered for relevant information, and an XML document of the results is written.