Related Work - User-Level Workflow Design: A Bioinformatics Perspective - page 175

that match the query. For each accession number, the corresponding sequence
is fetched from the DDBJ's DAD database (in FASTA format) and “blasted”
against the UniProt database. For each blast hit, the corresponding UniProt
entry is retrieved in order to extract the Protein ID and description. This
information is used to successively fill a table with accession numbers, protein
IDs, UniProt IDs, and protein descriptions. Thereby data from all scopes (not
just from the inner loop) is required for assembling the table data.
8.2.2 Control-Flow Realization
The Bio-jETI framework, as described in Chapter 2, was used in this study for
control-flow realizations of the benchmark workflows. The Bio-jETI version
that was used is based on jABC version 3.7 and jETI 1.3 (as released in
October 2008). As detailed before, the jABC is clearly a control-flow-oriented
environment, providing a set of different common control structure in the
form of SIBs, and expressing the control-flow via the connection between the
single SIBs, while the application data itself is managed within the associated
execution contexts.
Example 1 in Bio-jETI
The building blocks that are needed for assembling the multiple sequences
analysis in Bio-jETI are, on the one hand, SIBs that act as clients to the
involved EMBOSS services, and on the other hand SIBs that perform local
tasks like input handling and result representation. While a comprehensive
set of SIBs for common local tasks is provided by the jABC, the specific web
service clients have to be created first.
Fig. 8.4 Multiple sequences analysis workflow in Bio-jETI